chore: initial clean commit

.env

@@ -0,0 +1,22 @@
+# LLM Proxy Gateway Environment Variables
+
+# OpenAI
+OPENAI_API_KEY=sk-demo-openai-key
+
+# Google Gemini
+GEMINI_API_KEY=AIza-demo-gemini-key
+
+# DeepSeek
+DEEPSEEK_API_KEY=sk-demo-deepseek-key
+
+# xAI Grok (not yet available)
+GROK_API_KEY=gk-demo-grok-key
+
+# Authentication tokens (comma-separated list)
+LLM_PROXY__SERVER__AUTH_TOKENS=demo-token-123456,another-token
+
+# Server port (optional)
+LLM_PROXY__SERVER__PORT=8080
+
+# Database path (optional)
+LLM_PROXY__DATABASE__PATH=./data/llm_proxy.db

.env.example

@@ -0,0 +1,28 @@
+# LLM Proxy Gateway Environment Variables
+# Copy to .env and fill in your API keys
+
+# OpenAI
+OPENAI_API_KEY=your_openai_api_key_here
+
+# Google Gemini
+GEMINI_API_KEY=your_gemini_api_key_here
+
+# DeepSeek
+DEEPSEEK_API_KEY=your_deepseek_api_key_here
+
+# xAI Grok (not yet available)
+GROK_API_KEY=your_grok_api_key_here
+
+# Ollama (local server)
+# LLM_PROXY__PROVIDERS__OLLAMA__BASE_URL=http://your-ollama-host:11434/v1
+# LLM_PROXY__PROVIDERS__OLLAMA__ENABLED=true
+# LLM_PROXY__PROVIDERS__OLLAMA__MODELS=llama3,mistral,llava
+
+# Authentication tokens (comma-separated list)
+LLM_PROXY__SERVER__AUTH_TOKENS=your_bearer_token_here,another_token
+
+# Server port (optional)
+LLM_PROXY__SERVER__PORT=8080
+
+# Database path (optional)
+LLM_PROXY__DATABASE__PATH=./data/llm_proxy.db

.gitignore

@@ -0,0 +1,5 @@
+/target
+/.env
+/*.db
+/*.db-shm
+/*.db-wal

Cargo.toml

@@ -0,0 +1,75 @@
+[package]
+name = "llm-proxy"
+version = "0.1.0"
+edition = "2024"
+description = "Unified LLM proxy gateway supporting OpenAI, Gemini, DeepSeek, and Grok with token tracking and cost calculation"
+authors = ["newkirk"]
+license = "MIT OR Apache-2.0"
+repository = ""
+
+[dependencies]
+# ========== Web Framework & Async Runtime ==========
+axum = { version = "0.8", features = ["macros", "ws"] }
+tokio = { version = "1.0", features = ["rt-multi-thread", "macros", "net", "time", "signal", "fs"] }
+tower = "0.5"
+tower-http = { version = "0.6", features = ["trace", "cors", "compression-gzip", "fs"] }
+
+# ========== HTTP Clients ==========
+reqwest = { version = "0.12", default-features = false, features = ["json", "stream", "rustls-tls"] }
+async-openai = { version = "0.33", default-features = false, features = ["_api", "chat-completion"] }
+tiktoken-rs = "0.9"
+
+# ========== Database & ORM ==========
+sqlx = { version = "0.8", features = ["runtime-tokio", "sqlite", "macros", "migrate", "chrono"] }
+
+# ========== Authentication & Middleware ==========
+axum-extra = { version = "0.12", features = ["typed-header"] }
+headers = "0.4"
+
+# ========== Configuration Management ==========
+config = "0.13"
+dotenvy = "0.15"
+serde = { version = "1.0", features = ["derive"] }
+serde_json = "1.0"
+toml = "0.8"
+
+# ========== Logging & Monitoring ==========
+tracing = "0.1"
+tracing-subscriber = { version = "0.3", features = ["env-filter", "json"] }
+
+# ========== Multimodal & Image Processing ==========
+base64 = "0.21"
+image = { version = "0.25", default-features = false, features = ["jpeg", "png", "webp"] }
+mime = "0.3"
+mime_guess = "2.0"
+
+# ========== Error Handling & Utilities ==========
+anyhow = "1.0"
+thiserror = "1.0"
+chrono = { version = "0.4", features = ["serde"] }
+uuid = { version = "1.0", features = ["v4", "serde"] }
+futures = "0.3"
+async-trait = "0.1"
+async-stream = "0.3"
+reqwest-eventsource = "0.6"
+once_cell = "1.19"
+regex = "1.10"
+rand = "0.8"
+
+# ========== Rate Limiting & Circuit Breaking ==========
+governor = "0.6"
+
+[dev-dependencies]
+tokio-test = "0.4"
+mockito = "1.0"
+tempfile = "3.10"
+assert_cmd = "2.0"
+insta = "1.39"
+anyhow = "1.0"
+
+[profile.release]
+opt-level = 3
+lto = true
+codegen-units = 1
+strip = true
+panic = "abort"

DASHBOARD_README.md

@@ -0,0 +1,207 @@
+# LLM Proxy Gateway - Admin Dashboard
+
+## Overview
+
+This is a comprehensive admin dashboard for the LLM Proxy Gateway, providing real-time monitoring, analytics, and management capabilities for the proxy service.
+
+## Features
+
+### 1. **Dashboard Overview**
+- Real-time request counters and statistics
+- System health indicators
+- Provider status monitoring
+- Recent requests stream
+
+### 2. **Usage Analytics**
+- Time series charts for requests, tokens, and costs
+- Filter by date range, client, provider, and model
+- Top clients and models analysis
+- Export functionality to CSV/JSON
+
+### 3. **Cost Management**
+- Cost breakdown by provider, client, and model
+- Budget tracking with alerts
+- Cost projections
+- Pricing configuration management
+
+### 4. **Client Management**
+- List, create, revoke, and rotate API tokens
+- Client-specific rate limits
+- Usage statistics per client
+- Token management interface
+
+### 5. **Provider Configuration**
+- Enable/disable LLM providers
+- Configure API keys (masked display)
+- Test provider connections
+- Model availability management
+
+### 6. **Real-time Monitoring**
+- Live request stream via WebSocket
+- System metrics dashboard
+- Response time and error rate tracking
+- Live system logs
+
+### 7. **System Settings**
+- General configuration
+- Database management
+- Logging settings
+- Security settings
+
+## Technology Stack
+
+### Frontend
+- **HTML5/CSS3**: Modern, responsive design with CSS Grid/Flexbox
+- **JavaScript (ES6+)**: Vanilla JavaScript with modular architecture
+- **Chart.js**: Interactive data visualizations
+- **Luxon**: Date/time manipulation
+- **WebSocket API**: Real-time updates
+
+### Backend (Rust/Axum)
+- **Axum**: Web framework with WebSocket support
+- **Tokio**: Async runtime
+- **Serde**: JSON serialization/deserialization
+- **Broadcast channels**: Real-time event distribution
+
+## Installation & Setup
+
+### 1. Build and Run the Server
+```bash
+# Build the project
+cargo build --release
+
+# Run the server
+cargo run --release
+```
+
+### 2. Access the Dashboard
+Once the server is running, access the dashboard at:
+```
+http://localhost:8080
+```
+
+### 3. Default Login Credentials
+- **Username**: `admin`
+- **Password**: `admin123`
+
+## API Endpoints
+
+### Authentication
+- `POST /api/auth/login` - Dashboard login
+- `GET /api/auth/status` - Authentication status
+
+### Analytics
+- `GET /api/usage/summary` - Overall usage summary
+- `GET /api/usage/time-series` - Time series data
+- `GET /api/usage/clients` - Client breakdown
+- `GET /api/usage/providers` - Provider breakdown
+
+### Clients
+- `GET /api/clients` - List all clients
+- `POST /api/clients` - Create new client
+- `DELETE /api/clients/{id}` - Revoke client
+- `GET /api/clients/{id}/usage` - Client-specific usage
+
+### Providers
+- `GET /api/providers` - List providers and status
+- `PUT /api/providers/{name}` - Update provider config
+- `POST /api/providers/{name}/test` - Test provider connection
+
+### System
+- `GET /api/system/health` - System health
+- `GET /api/system/logs` - Recent logs
+- `POST /api/system/backup` - Trigger backup
+
+### WebSocket
+- `GET /ws` - WebSocket endpoint for real-time updates
+
+## Project Structure
+
+```
+llm-proxy/
+├── src/
+│   ├── dashboard/          # Dashboard backend module
+│   │   └── mod.rs         # Dashboard routes and handlers
+│   ├── server/            # Main proxy server
+│   ├── providers/         # LLM provider implementations
+│   └── ...               # Other modules
+├── static/               # Frontend dashboard files
+│   ├── index.html        # Main dashboard HTML
+│   ├── css/
+│   │   └── dashboard.css # Dashboard styles
+│   ├── js/
+│   │   ├── auth.js       # Authentication module
+│   │   ├── dashboard.js  # Main dashboard controller
+│   │   ├── websocket.js  # WebSocket manager
+│   │   ├── charts.js     # Chart.js utilities
+│   │   └── pages/        # Page-specific modules
+│   │       ├── overview.js
+│   │       ├── analytics.js
+│   │       ├── costs.js
+│   │       ├── clients.js
+│   │       ├── providers.js
+│   │       ├── monitoring.js
+│   │       ├── settings.js
+│   │       └── logs.js
+│   ├── img/              # Images and icons
+│   └── fonts/            # Font files
+└── Cargo.toml           # Rust dependencies
+```
+
+## Development
+
+### Adding New Pages
+1. Create a new JavaScript module in `static/js/pages/`
+2. Implement the page class with `init()` method
+3. Register the page in `dashboard.js`
+4. Add menu item in `index.html`
+
+### Adding New API Endpoints
+1. Add route in `src/dashboard/mod.rs`
+2. Implement handler function
+3. Update frontend JavaScript to call the endpoint
+
+### Styling Guidelines
+- Use CSS custom properties (variables) from `:root`
+- Follow mobile-first responsive design
+- Use BEM-like naming convention for CSS classes
+- Maintain consistent spacing with CSS variables
+
+## Security Considerations
+
+1. **Authentication**: Simple password-based auth for demo; replace with proper auth in production
+2. **API Keys**: Tokens are masked in the UI (only last 4 characters shown)
+3. **CORS**: Configure appropriate CORS headers for production
+4. **Rate Limiting**: Implement rate limiting for API endpoints
+5. **HTTPS**: Always use HTTPS in production
+
+## Performance Optimizations
+
+1. **Code Splitting**: JavaScript modules are loaded on-demand
+2. **Caching**: Static assets are served with cache headers
+3. **WebSocket**: Real-time updates reduce polling overhead
+4. **Lazy Loading**: Charts and tables load data as needed
+5. **Compression**: Enable gzip/brotli compression for static files
+
+## Browser Support
+
+- Chrome 60+
+- Firefox 55+
+- Safari 11+
+- Edge 79+
+
+## License
+
+MIT License - See LICENSE file for details.
+
+## Contributing
+
+1. Fork the repository
+2. Create a feature branch
+3. Make your changes
+4. Add tests if applicable
+5. Submit a pull request
+
+## Support
+
+For issues and feature requests, please use the GitHub issue tracker.

OPTIMIZATION.md

@@ -0,0 +1,232 @@
+# Optimization for 512MB RAM Environment
+
+This document provides guidance for optimizing the LLM Proxy Gateway for deployment in resource-constrained environments (512MB RAM).
+
+## Memory Optimization Strategies
+
+### 1. Build Optimization
+
+The project is already configured with optimized build settings in `Cargo.toml`:
+
+```toml
+[profile.release]
+opt-level = 3      # Maximum optimization
+lto = true         # Link-time optimization
+codegen-units = 1  # Single codegen unit for better optimization
+strip = true       # Strip debug symbols
+```
+
+**Additional optimizations you can apply:**
+
+```bash
+# Build with specific target for better optimization
+cargo build --release --target x86_64-unknown-linux-musl
+
+# Or for ARM (Raspberry Pi, etc.)
+cargo build --release --target aarch64-unknown-linux-musl
+```
+
+### 2. Runtime Memory Management
+
+#### Database Connection Pool
+- Default: 10 connections
+- Recommended for 512MB: 5 connections
+
+Update `config.toml`:
+```toml
+[database]
+max_connections = 5
+```
+
+#### Rate Limiting Memory Usage
+- Client rate limit buckets: Store in memory
+- Circuit breakers: Minimal memory usage
+- Consider reducing burst capacity if memory is critical
+
+#### Provider Management
+- Only enable providers you actually use
+- Disable unused providers in configuration
+
+### 3. Configuration for Low Memory
+
+Create a `config-low-memory.toml`:
+
+```toml
+[server]
+port = 8080
+host = "0.0.0.0"
+
+[database]
+path = "./data/llm_proxy.db"
+max_connections = 3  # Reduced from default 10
+
+[providers]
+# Only enable providers you need
+openai.enabled = true
+gemini.enabled = false  # Disable if not used
+deepseek.enabled = false  # Disable if not used
+grok.enabled = false  # Disable if not used
+
+[rate_limiting]
+# Reduce memory usage for rate limiting
+client_requests_per_minute = 30  # Reduced from 60
+client_burst_size = 5  # Reduced from 10
+global_requests_per_minute = 300  # Reduced from 600
+```
+
+### 4. System-Level Optimizations
+
+#### Linux Kernel Parameters
+Add to `/etc/sysctl.conf`:
+```bash
+# Reduce TCP buffer sizes
+net.ipv4.tcp_rmem = 4096 87380 174760
+net.ipv4.tcp_wmem = 4096 65536 131072
+
+# Reduce connection tracking
+net.netfilter.nf_conntrack_max = 65536
+net.netfilter.nf_conntrack_tcp_timeout_established = 1200
+
+# Reduce socket buffer sizes
+net.core.rmem_max = 131072
+net.core.wmem_max = 131072
+net.core.rmem_default = 65536
+net.core.wmem_default = 65536
+```
+
+#### Systemd Service Configuration
+Create `/etc/systemd/system/llm-proxy.service`:
+```ini
+[Unit]
+Description=LLM Proxy Gateway
+After=network.target
+
+[Service]
+Type=simple
+User=llmproxy
+Group=llmproxy
+WorkingDirectory=/opt/llm-proxy
+ExecStart=/opt/llm-proxy/llm-proxy
+Restart=on-failure
+RestartSec=5
+
+# Memory limits
+MemoryMax=400M
+MemorySwapMax=100M
+
+# CPU limits
+CPUQuota=50%
+
+# Process limits
+LimitNOFILE=65536
+LimitNPROC=512
+
+Environment="RUST_LOG=info"
+Environment="LLM_PROXY__DATABASE__MAX_CONNECTIONS=3"
+
+[Install]
+WantedBy=multi-user.target
+```
+
+### 5. Application-Specific Optimizations
+
+#### Disable Unused Features
+- **Multimodal support**: If not using images, disable image processing dependencies
+- **Dashboard**: The dashboard uses WebSockets and additional memory. Consider disabling if not needed.
+- **Detailed logging**: Reduce log verbosity in production
+
+#### Memory Pool Sizes
+The application uses several memory pools:
+1. **Database connection pool**: Configured via `max_connections`
+2. **HTTP client pool**: Reqwest client pool (defaults to reasonable values)
+3. **Async runtime**: Tokio worker threads
+
+Reduce Tokio worker threads for low-core systems:
+```rust
+// In main.rs, modify tokio runtime creation
+#[tokio::main(flavor = "current_thread")]  // Single-threaded runtime
+async fn main() -> Result<()> {
+    // Or for multi-threaded with limited threads:
+    // #[tokio::main(worker_threads = 2)]
+```
+
+### 6. Monitoring and Profiling
+
+#### Memory Usage Monitoring
+```bash
+# Install heaptrack for memory profiling
+cargo install heaptrack
+
+# Profile memory usage
+heaptrack ./target/release/llm-proxy
+
+# Monitor with ps
+ps aux --sort=-%mem | head -10
+
+# Monitor with top
+top -p $(pgrep llm-proxy)
+```
+
+#### Performance Benchmarks
+Test with different configurations:
+```bash
+# Test with 100 concurrent connections
+wrk -t4 -c100 -d30s http://localhost:8080/health
+
+# Test chat completion endpoint
+ab -n 1000 -c 10 -p test_request.json -T application/json http://localhost:8080/v1/chat/completions
+```
+
+### 7. Deployment Checklist for 512MB RAM
+
+- [ ] Build with release profile: `cargo build --release`
+- [ ] Configure database with `max_connections = 3`
+- [ ] Disable unused providers in configuration
+- [ ] Set appropriate rate limiting limits
+- [ ] Configure systemd with memory limits
+- [ ] Set up log rotation to prevent disk space issues
+- [ ] Monitor memory usage during initial deployment
+- [ ] Consider using swap space (512MB-1GB) for safety
+
+### 8. Troubleshooting High Memory Usage
+
+#### Common Issues and Solutions:
+
+1. **Database connection leaks**: Ensure connections are properly closed
+2. **Memory fragmentation**: Use jemalloc or mimalloc as allocator
+3. **Unbounded queues**: Check WebSocket message queues
+4. **Cache growth**: Implement cache limits or TTL
+
+#### Add to Cargo.toml for alternative allocator:
+```toml
+[dependencies]
+mimalloc = { version = "0.1", default-features = false }
+
+[features]
+default = ["mimalloc"]
+```
+
+#### In main.rs:
+```rust
+#[global_allocator]
+static GLOBAL: mimalloc::MiMalloc = mimalloc::MiMalloc;
+```
+
+### 9. Expected Memory Usage
+
+| Component | Baseline | With 10 clients | With 100 clients |
+|-----------|----------|-----------------|------------------|
+| Base executable | 15MB | 15MB | 15MB |
+| Database connections | 5MB | 8MB | 15MB |
+| Rate limiting | 2MB | 5MB | 20MB |
+| HTTP clients | 3MB | 5MB | 10MB |
+| **Total** | **25MB** | **33MB** | **60MB** |
+
+**Note**: These are estimates. Actual usage depends on request volume, payload sizes, and configuration.
+
+### 10. Further Reading
+
+- [Tokio performance guide](https://tokio.rs/tokio/topics/performance)
+- [Rust performance book](https://nnethercote.github.io/perf-book/)
+- [Linux memory management](https://www.kernel.org/doc/html/latest/admin-guide/mm/)
+- [SQLite performance tips](https://www.sqlite.org/faq.html#q19)

README.md

@@ -0,0 +1,100 @@
+# LLM Proxy Gateway
+
+A unified, high-performance LLM proxy gateway built in Rust. It provides a single OpenAI-compatible API to access multiple providers (OpenAI, Gemini, DeepSeek, Grok) with built-in token tracking, real-time cost calculation, and a management dashboard.
+
+## 🚀 Features
+
+-   **Unified API:** Fully OpenAI-compatible `/v1/chat/completions` endpoint.
+-   **Multi-Provider Support:**
+    *   **OpenAI:** Standard models (GPT-4o, GPT-3.5, etc.) and reasoning models (o1, o3).
+    *   **Google Gemini:** Support for the latest Gemini 2.0 models.
+    *   **DeepSeek:** High-performance, low-cost integration.
+    *   **xAI Grok:** Integration for Grok-series models.
+    *   **Ollama:** Support for local LLMs running on your machine or another host.
+-   **Observability & Tracking:**
+    *   **Real-time Costing:** Fetches live pricing and context specs from `models.dev` on startup.
+    *   **Token Counting:** Precise estimation using `tiktoken-rs`.
+    *   **Database Logging:** Every request is logged to SQLite for historical analysis.
+    *   **Streaming Support:** Full SSE (Server-Sent Events) support with aggregated token tracking.
+-   **Multimodal (Vision):** Support for image processing (Base64 and remote URLs) across compatible providers.
+-   **Reliability:**
+    *   **Circuit Breaking:** Automatically protects your system when providers are down.
+    *   **Rate Limiting:** Granular per-client and global rate limits.
+-   **Management Dashboard:** A modern, real-time web interface to monitor usage, costs, and system health.
+
+## 🛠️ Tech Stack
+
+-   **Runtime:** Rust (2024 Edition) with [Tokio](https://tokio.rs/).
+-   **Web Framework:** [Axum](https://github.com/tokio-rs/axum).
+-   **Database:** [SQLx](https://github.com/launchbadge/sqlx) with SQLite.
+-   **Frontend:** Vanilla JS/CSS (no heavyweight framework required).
+
+## 🚦 Getting Started
+
+### Prerequisites
+
+-   Rust (1.80+)
+-   SQLite3
+
+### Installation
+
+1.  Clone the repository:
+    ```bash
+    git clone https://github.com/yourusername/llm-proxy.git
+    cd llm-proxy
+    ```
+
+2.  Set up your environment:
+    ```bash
+    cp .env.example .env
+    # Edit .env and add your API keys
+    ```
+
+3.  Configure providers and server:
+    Edit `config.toml` to customize models, pricing fallbacks, and port settings.
+
+    **Ollama Example (config.toml):**
+    ```toml
+    [providers.ollama]
+    enabled = true
+    base_url = "http://192.168.1.50:11434/v1"
+    models = ["llama3", "mistral"]
+    ```
+
+4.  Run the proxy:
+    ```bash
+    cargo run --release
+    ```
+
+The server will start at `http://localhost:3000` (by default).
+
+## 📊 Management Dashboard
+
+Access the built-in dashboard at `http://localhost:3000` to see:
+-   **Usage Summary:** Total requests, tokens, and USD spent.
+-   **Trend Charts:** 24-hour request and cost distributions.
+-   **Live Logs:** Real-time stream of incoming LLM requests via WebSockets.
+-   **Provider Health:** Monitor which providers are online or degraded.
+
+## 🔌 API Usage
+
+The proxy is designed to be a drop-in replacement for OpenAI. Simply change your base URL:
+
+**Example Request (Python):**
+```python
+from openai import OpenAI
+
+client = OpenAI(
+    base_url="http://localhost:3000/v1",
+    api_key="your-proxy-client-id" # Hashed sk- keys are managed in the dashboard
+)
+
+response = client.chat.completions.create(
+    model="gpt-4o",
+    messages=[{"role": "user", "content": "Hello from the proxy!"}]
+)
+```
+
+## ⚖️ License
+
+MIT OR Apache-2.0

deploy.sh

@@ -0,0 +1,610 @@
+#!/bin/bash
+
+# LLM Proxy Gateway Deployment Script
+# This script automates the deployment of the LLM Proxy Gateway on a Linux server
+
+set -e  # Exit on error
+set -u  # Exit on undefined variable
+
+# Configuration
+APP_NAME="llm-proxy"
+APP_USER="llmproxy"
+APP_GROUP="llmproxy"
+INSTALL_DIR="/opt/$APP_NAME"
+CONFIG_DIR="/etc/$APP_NAME"
+DATA_DIR="/var/lib/$APP_NAME"
+LOG_DIR="/var/log/$APP_NAME"
+SERVICE_FILE="/etc/systemd/system/$APP_NAME.service"
+ENV_FILE="$CONFIG_DIR/.env"
+
+# Colors for output
+RED='\033[0;31m'
+GREEN='\033[0;32m'
+YELLOW='\033[1;33m'
+NC='\033[0m' # No Color
+
+# Logging functions
+log_info() {
+    echo -e "${GREEN}[INFO]${NC} $1"
+}
+
+log_warn() {
+    echo -e "${YELLOW}[WARN]${NC} $1"
+}
+
+log_error() {
+    echo -e "${RED}[ERROR]${NC} $1"
+}
+
+# Check if running as root
+check_root() {
+    if [[ $EUID -ne 0 ]]; then
+        log_error "This script must be run as root"
+        exit 1
+    fi
+}
+
+# Install system dependencies
+install_dependencies() {
+    log_info "Installing system dependencies..."
+    
+    # Detect package manager
+    if command -v apt-get &> /dev/null; then
+        # Debian/Ubuntu
+        apt-get update
+        apt-get install -y \
+            build-essential \
+            pkg-config \
+            libssl-dev \
+            sqlite3 \
+            curl \
+            git
+    elif command -v yum &> /dev/null; then
+        # RHEL/CentOS
+        yum groupinstall -y "Development Tools"
+        yum install -y \
+            openssl-devel \
+            sqlite \
+            curl \
+            git
+    elif command -v dnf &> /dev/null; then
+        # Fedora
+        dnf groupinstall -y "Development Tools"
+        dnf install -y \
+            openssl-devel \
+            sqlite \
+            curl \
+            git
+    elif command -v pacman &> /dev/null; then
+        # Arch Linux
+        pacman -Syu --noconfirm \
+            base-devel \
+            openssl \
+            sqlite \
+            curl \
+            git
+    else
+        log_warn "Could not detect package manager. Please install dependencies manually."
+    fi
+}
+
+# Install Rust if not present
+install_rust() {
+    log_info "Checking for Rust installation..."
+    
+    if ! command -v rustc &> /dev/null; then
+        log_info "Installing Rust..."
+        curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+        source "$HOME/.cargo/env"
+    else
+        log_info "Rust is already installed"
+    fi
+    
+    # Verify installation
+    rustc --version
+    cargo --version
+}
+
+# Create system user and directories
+setup_directories() {
+    log_info "Creating system user and directories..."
+    
+    # Create user and group if they don't exist
+    if ! id "$APP_USER" &>/dev/null; then
+        useradd -r -s /usr/sbin/nologin -M "$APP_USER"
+    fi
+    
+    # Create directories
+    mkdir -p "$INSTALL_DIR"
+    mkdir -p "$CONFIG_DIR"
+    mkdir -p "$DATA_DIR"
+    mkdir -p "$LOG_DIR"
+    
+    # Set permissions
+    chown -R "$APP_USER:$APP_GROUP" "$INSTALL_DIR"
+    chown -R "$APP_USER:$APP_GROUP" "$CONFIG_DIR"
+    chown -R "$APP_USER:$APP_GROUP" "$DATA_DIR"
+    chown -R "$APP_USER:$APP_GROUP" "$LOG_DIR"
+    
+    chmod 750 "$INSTALL_DIR"
+    chmod 750 "$CONFIG_DIR"
+    chmod 750 "$DATA_DIR"
+    chmod 750 "$LOG_DIR"
+}
+
+# Build the application
+build_application() {
+    log_info "Building the application..."
+    
+    # Clone or update repository
+    if [[ ! -d "$INSTALL_DIR/.git" ]]; then
+        log_info "Cloning repository..."
+        git clone https://github.com/yourusername/llm-proxy.git "$INSTALL_DIR"
+    else
+        log_info "Updating repository..."
+        cd "$INSTALL_DIR"
+        git pull
+    fi
+    
+    # Build in release mode
+    cd "$INSTALL_DIR"
+    log_info "Building release binary..."
+    cargo build --release
+    
+    # Verify build
+    if [[ -f "target/release/$APP_NAME" ]]; then
+        log_info "Build successful"
+    else
+        log_error "Build failed"
+        exit 1
+    fi
+}
+
+# Create configuration files
+create_configuration() {
+    log_info "Creating configuration files..."
+    
+    # Create .env file with API keys
+    cat > "$ENV_FILE" << EOF
+# LLM Proxy Gateway Environment Variables
+# Add your API keys here
+
+# OpenAI API Key
+# OPENAI_API_KEY=sk-your-key-here
+
+# Google Gemini API Key
+# GEMINI_API_KEY=AIza-your-key-here
+
+# DeepSeek API Key
+# DEEPSEEK_API_KEY=sk-your-key-here
+
+# xAI Grok API Key
+# GROK_API_KEY=gk-your-key-here
+
+# Authentication tokens (comma-separated)
+# LLM_PROXY__SERVER__AUTH_TOKENS=token1,token2,token3
+EOF
+    
+    # Create config.toml
+    cat > "$CONFIG_DIR/config.toml" << EOF
+# LLM Proxy Gateway Configuration
+
+[server]
+port = 8080
+host = "0.0.0.0"
+# auth_tokens = ["token1", "token2", "token3"]  # Uncomment to enable authentication
+
+[database]
+path = "$DATA_DIR/llm_proxy.db"
+max_connections = 5
+
+[providers.openai]
+enabled = true
+api_key_env = "OPENAI_API_KEY"
+base_url = "https://api.openai.com/v1"
+default_model = "gpt-4o"
+
+[providers.gemini]
+enabled = true
+api_key_env = "GEMINI_API_KEY"
+base_url = "https://generativelanguage.googleapis.com/v1"
+default_model = "gemini-2.0-flash"
+
+[providers.deepseek]
+enabled = true
+api_key_env = "DEEPSEEK_API_KEY"
+base_url = "https://api.deepseek.com"
+default_model = "deepseek-reasoner"
+
+[providers.grok]
+enabled = false  # Disabled by default until API is researched
+api_key_env = "GROK_API_KEY"
+base_url = "https://api.x.ai/v1"
+default_model = "grok-beta"
+
+[model_mapping]
+"gpt-*" = "openai"
+"gemini-*" = "gemini"
+"deepseek-*" = "deepseek"
+"grok-*" = "grok"
+
+[pricing]
+openai = { input = 0.01, output = 0.03 }
+gemini = { input = 0.0005, output = 0.0015 }
+deepseek = { input = 0.00014, output = 0.00028 }
+grok = { input = 0.001, output = 0.003 }
+EOF
+    
+    # Set permissions
+    chown "$APP_USER:$APP_GROUP" "$ENV_FILE"
+    chown "$APP_USER:$APP_GROUP" "$CONFIG_DIR/config.toml"
+    chmod 640 "$ENV_FILE"
+    chmod 640 "$CONFIG_DIR/config.toml"
+}
+
+# Create systemd service
+create_systemd_service() {
+    log_info "Creating systemd service..."
+    
+    cat > "$SERVICE_FILE" << EOF
+[Unit]
+Description=LLM Proxy Gateway
+Documentation=https://github.com/yourusername/llm-proxy
+After=network.target
+Wants=network.target
+
+[Service]
+Type=simple
+User=$APP_USER
+Group=$APP_GROUP
+WorkingDirectory=$INSTALL_DIR
+EnvironmentFile=$ENV_FILE
+Environment="RUST_LOG=info"
+Environment="LLM_PROXY__CONFIG_PATH=$CONFIG_DIR/config.toml"
+ExecStart=$INSTALL_DIR/target/release/$APP_NAME
+Restart=on-failure
+RestartSec=5
+
+# Security hardening
+NoNewPrivileges=true
+PrivateTmp=true
+ProtectSystem=strict
+ProtectHome=true
+ReadWritePaths=$DATA_DIR $LOG_DIR
+
+# Resource limits (adjust based on your server)
+MemoryMax=400M
+MemorySwapMax=100M
+CPUQuota=50%
+LimitNOFILE=65536
+
+# Logging
+StandardOutput=journal
+StandardError=journal
+SyslogIdentifier=$APP_NAME
+
+[Install]
+WantedBy=multi-user.target
+EOF
+    
+    # Reload systemd
+    systemctl daemon-reload
+}
+
+# Setup nginx reverse proxy (optional)
+setup_nginx_proxy() {
+    if ! command -v nginx &> /dev/null; then
+        log_warn "nginx not installed. Skipping reverse proxy setup."
+        return
+    fi
+    
+    log_info "Setting up nginx reverse proxy..."
+    
+    cat > "/etc/nginx/sites-available/$APP_NAME" << EOF
+server {
+    listen 80;
+    server_name your-domain.com;  # Change to your domain
+    
+    # Redirect to HTTPS (recommended)
+    return 301 https://\$server_name\$request_uri;
+}
+
+server {
+    listen 443 ssl http2;
+    server_name your-domain.com;  # Change to your domain
+    
+    # SSL certificates (adjust paths)
+    ssl_certificate /etc/letsencrypt/live/your-domain.com/fullchain.pem;
+    ssl_certificate_key /etc/letsencrypt/live/your-domain.com/privkey.pem;
+    
+    # SSL configuration
+    ssl_protocols TLSv1.2 TLSv1.3;
+    ssl_ciphers ECDHE-RSA-AES256-GCM-SHA512:DHE-RSA-AES256-GCM-SHA512:ECDHE-RSA-AES256-GCM-SHA384:DHE-RSA-AES256-GCM-SHA384;
+    ssl_prefer_server_ciphers off;
+    
+    # Proxy to LLM Proxy Gateway
+    location / {
+        proxy_pass http://127.0.0.1:8080;
+        proxy_http_version 1.1;
+        proxy_set_header Upgrade \$http_upgrade;
+        proxy_set_header Connection "upgrade";
+        proxy_set_header Host \$host;
+        proxy_set_header X-Real-IP \$remote_addr;
+        proxy_set_header X-Forwarded-For \$proxy_add_x_forwarded_for;
+        proxy_set_header X-Forwarded-Proto \$scheme;
+        
+        # Timeouts
+        proxy_connect_timeout 60s;
+        proxy_send_timeout 60s;
+        proxy_read_timeout 60s;
+    }
+    
+    # Health check endpoint
+    location /health {
+        proxy_pass http://127.0.0.1:8080/health;
+        access_log off;
+    }
+    
+    # Dashboard
+    location /dashboard {
+        proxy_pass http://127.0.0.1:8080/dashboard;
+    }
+}
+EOF
+    
+    # Enable site
+    ln -sf "/etc/nginx/sites-available/$APP_NAME" "/etc/nginx/sites-enabled/"
+    
+    # Test nginx configuration
+    nginx -t
+    
+    log_info "nginx configuration created. Please update the domain and SSL certificate paths."
+}
+
+# Setup firewall
+setup_firewall() {
+    log_info "Configuring firewall..."
+    
+    # Check for ufw (Ubuntu)
+    if command -v ufw &> /dev/null; then
+        ufw allow 22/tcp  # SSH
+        ufw allow 80/tcp  # HTTP
+        ufw allow 443/tcp # HTTPS
+        ufw --force enable
+        log_info "UFW firewall configured"
+    fi
+    
+    # Check for firewalld (RHEL/CentOS)
+    if command -v firewall-cmd &> /dev/null; then
+        firewall-cmd --permanent --add-service=ssh
+        firewall-cmd --permanent --add-service=http
+        firewall-cmd --permanent --add-service=https
+        firewall-cmd --reload
+        log_info "Firewalld configured"
+    fi
+}
+
+# Initialize database
+initialize_database() {
+    log_info "Initializing database..."
+    
+    # Run the application once to create database
+    sudo -u "$APP_USER" "$INSTALL_DIR/target/release/$APP_NAME" --help &> /dev/null || true
+    
+    log_info "Database initialized at $DATA_DIR/llm_proxy.db"
+}
+
+# Start and enable service
+start_service() {
+    log_info "Starting $APP_NAME service..."
+    
+    systemctl enable "$APP_NAME"
+    systemctl start "$APP_NAME"
+    
+    # Check status
+    sleep 2
+    systemctl status "$APP_NAME" --no-pager
+}
+
+# Verify installation
+verify_installation() {
+    log_info "Verifying installation..."
+    
+    # Check if service is running
+    if systemctl is-active --quiet "$APP_NAME"; then
+        log_info "Service is running"
+    else
+        log_error "Service is not running"
+        journalctl -u "$APP_NAME" -n 20 --no-pager
+        exit 1
+    fi
+    
+    # Test health endpoint
+    if curl -s http://localhost:8080/health | grep -q "OK"; then
+        log_info "Health check passed"
+    else
+        log_error "Health check failed"
+        exit 1
+    fi
+    
+    # Test dashboard
+    if curl -s -o /dev/null -w "%{http_code}" http://localhost:8080/dashboard | grep -q "200"; then
+        log_info "Dashboard is accessible"
+    else
+        log_warn "Dashboard may not be accessible (this is normal if not configured)"
+    fi
+    
+    log_info "Installation verified successfully!"
+}
+
+# Print next steps
+print_next_steps() {
+    cat << EOF
+
+${GREEN}=== LLM Proxy Gateway Installation Complete ===${NC}
+
+${YELLOW}Next steps:${NC}
+
+1. ${GREEN}Configure API keys${NC}
+   Edit: $ENV_FILE
+   Add your API keys for the providers you want to use
+
+2. ${GREEN}Configure authentication${NC}
+   Edit: $CONFIG_DIR/config.toml
+   Uncomment and set auth_tokens for client authentication
+
+3. ${GREEN}Configure nginx${NC}
+   Edit: /etc/nginx/sites-available/$APP_NAME
+   Update domain name and SSL certificate paths
+
+4. ${GREEN}Test the API${NC}
+   curl -X POST http://localhost:8080/v1/chat/completions \\
+     -H "Content-Type: application/json" \\
+     -H "Authorization: Bearer your-token" \\
+     -d '{
+       "model": "gpt-4o",
+       "messages": [{"role": "user", "content": "Hello!"}]
+     }'
+
+5. ${GREEN}Access the dashboard${NC}
+   Open: http://your-server-ip:8080/dashboard
+   Or: https://your-domain.com/dashboard (if nginx configured)
+
+${YELLOW}Useful commands:${NC}
+   systemctl status $APP_NAME    # Check service status
+   journalctl -u $APP_NAME -f    # View logs
+   systemctl restart $APP_NAME   # Restart service
+
+${YELLOW}Configuration files:${NC}
+   Service: $SERVICE_FILE
+   Config: $CONFIG_DIR/config.toml
+   Environment: $ENV_FILE
+   Database: $DATA_DIR/llm_proxy.db
+   Logs: $LOG_DIR/
+
+${GREEN}For more information, see:${NC}
+   https://github.com/yourusername/llm-proxy
+   $INSTALL_DIR/README.md
+   $INSTALL_DIR/deployment.md
+
+EOF
+}
+
+# Main deployment function
+deploy() {
+    log_info "Starting LLM Proxy Gateway deployment..."
+    
+    check_root
+    install_dependencies
+    install_rust
+    setup_directories
+    build_application
+    create_configuration
+    create_systemd_service
+    initialize_database
+    start_service
+    verify_installation
+    print_next_steps
+    
+    # Optional steps (uncomment if needed)
+    # setup_nginx_proxy
+    # setup_firewall
+    
+    log_info "Deployment completed successfully!"
+}
+
+# Update function
+update() {
+    log_info "Updating LLM Proxy Gateway..."
+    
+    check_root
+    
+    # Stop service
+    systemctl stop "$APP_NAME"
+    
+    # Update from git
+    cd "$INSTALL_DIR"
+    git pull
+    
+    # Rebuild
+    cargo build --release
+    
+    # Restart service
+    systemctl start "$APP_NAME"
+    
+    log_info "Update completed successfully!"
+    systemctl status "$APP_NAME" --no-pager
+}
+
+# Uninstall function
+uninstall() {
+    log_info "Uninstalling LLM Proxy Gateway..."
+    
+    check_root
+    
+    # Stop and disable service
+    systemctl stop "$APP_NAME" 2>/dev/null || true
+    systemctl disable "$APP_NAME" 2>/dev/null || true
+    rm -f "$SERVICE_FILE"
+    systemctl daemon-reload
+    
+    # Remove application files
+    rm -rf "$INSTALL_DIR"
+    rm -rf "$CONFIG_DIR"
+    
+    # Keep data and logs (comment out to remove)
+    log_warn "Data directory $DATA_DIR and logs $LOG_DIR have been preserved"
+    log_warn "Remove manually if desired:"
+    log_warn "  rm -rf $DATA_DIR $LOG_DIR"
+    
+    # Remove user (optional)
+    read -p "Remove user $APP_USER? [y/N]: " -n 1 -r
+    echo
+    if [[ $REPLY =~ ^[Yy]$ ]]; then
+        userdel "$APP_USER" 2>/dev/null || true
+        groupdel "$APP_GROUP" 2>/dev/null || true
+    fi
+    
+    log_info "Uninstallation completed!"
+}
+
+# Show usage
+usage() {
+    cat << EOF
+LLM Proxy Gateway Deployment Script
+
+Usage: $0 [command]
+
+Commands:
+  deploy     - Install and configure LLM Proxy Gateway
+  update     - Update existing installation
+  uninstall  - Remove LLM Proxy Gateway
+  help       - Show this help message
+
+Examples:
+  $0 deploy     # Full installation
+  $0 update     # Update to latest version
+  $0 uninstall  # Remove installation
+
+EOF
+}
+
+# Parse command line arguments
+case "${1:-}" in
+    deploy)
+        deploy
+        ;;
+    update)
+        update
+        ;;
+    uninstall)
+        uninstall
+        ;;
+    help|--help|-h)
+        usage
+        ;;
+    *)
+        usage
+        exit 1
+        ;;
+esac

deployment.md

@@ -0,0 +1,294 @@
+# LLM Proxy Gateway - Deployment Guide
+
+## Overview
+A unified LLM proxy gateway supporting OpenAI, Google Gemini, DeepSeek, and xAI Grok with token tracking, cost calculation, and admin dashboard.
+
+## System Requirements
+- **CPU**: 2 cores minimum
+- **RAM**: 512MB minimum (1GB recommended)
+- **Storage**: 10GB minimum
+- **OS**: Linux (tested on Arch Linux, Ubuntu, Debian)
+- **Runtime**: Rust 1.70+ with Cargo
+
+## Deployment Options
+
+### Option 1: Docker (Recommended)
+```dockerfile
+FROM rust:1.70-alpine as builder
+WORKDIR /app
+COPY . .
+RUN cargo build --release
+
+FROM alpine:latest
+RUN apk add --no-cache libgcc
+COPY --from=builder /app/target/release/llm-proxy /usr/local/bin/
+COPY --from=builder /app/static /app/static
+WORKDIR /app
+EXPOSE 8080
+CMD ["llm-proxy"]
+```
+
+### Option 2: Systemd Service (Bare Metal/LXC)
+```ini
+# /etc/systemd/system/llm-proxy.service
+[Unit]
+Description=LLM Proxy Gateway
+After=network.target
+
+[Service]
+Type=simple
+User=llmproxy
+Group=llmproxy
+WorkingDirectory=/opt/llm-proxy
+ExecStart=/opt/llm-proxy/llm-proxy
+Restart=always
+RestartSec=10
+Environment="RUST_LOG=info"
+Environment="LLM_PROXY__SERVER__PORT=8080"
+Environment="LLM_PROXY__SERVER__AUTH_TOKENS=sk-test-123,sk-test-456"
+
+[Install]
+WantedBy=multi-user.target
+```
+
+### Option 3: LXC Container (Proxmox)
+1. Create Alpine Linux LXC container
+2. Install Rust: `apk add rust cargo`
+3. Copy application files
+4. Build: `cargo build --release`
+5. Run: `./target/release/llm-proxy`
+
+## Configuration
+
+### Environment Variables
+```bash
+# Required API Keys
+OPENAI_API_KEY=sk-...
+GEMINI_API_KEY=AIza...
+DEEPSEEK_API_KEY=sk-...
+GROK_API_KEY=gk-...  # Optional
+
+# Server Configuration (with LLM_PROXY__ prefix)
+LLM_PROXY__SERVER__PORT=8080
+LLM_PROXY__SERVER__HOST=0.0.0.0
+LLM_PROXY__SERVER__AUTH_TOKENS=sk-test-123,sk-test-456
+
+# Database Configuration
+LLM_PROXY__DATABASE__PATH=./data/llm_proxy.db
+LLM_PROXY__DATABASE__MAX_CONNECTIONS=10
+
+# Provider Configuration
+LLM_PROXY__PROVIDERS__OPENAI__ENABLED=true
+LLM_PROXY__PROVIDERS__GEMINI__ENABLED=true
+LLM_PROXY__PROVIDERS__DEEPSEEK__ENABLED=true
+LLM_PROXY__PROVIDERS__GROK__ENABLED=false
+```
+
+### Configuration File (config.toml)
+Create `config.toml` in the application directory:
+```toml
+[server]
+port = 8080
+host = "0.0.0.0"
+auth_tokens = ["sk-test-123", "sk-test-456"]
+
+[database]
+path = "./data/llm_proxy.db"
+max_connections = 10
+
+[providers.openai]
+enabled = true
+base_url = "https://api.openai.com/v1"
+default_model = "gpt-4o"
+
+[providers.gemini]
+enabled = true
+base_url = "https://generativelanguage.googleapis.com/v1"
+default_model = "gemini-2.0-flash"
+
+[providers.deepseek]
+enabled = true
+base_url = "https://api.deepseek.com"
+default_model = "deepseek-reasoner"
+
+[providers.grok]
+enabled = false
+base_url = "https://api.x.ai/v1"
+default_model = "grok-beta"
+```
+
+## Nginx Reverse Proxy Configuration
+```nginx
+server {
+    listen 80;
+    server_name llm-proxy.yourdomain.com;
+    
+    location / {
+        proxy_pass http://localhost:8080;
+        proxy_http_version 1.1;
+        proxy_set_header Upgrade $http_upgrade;
+        proxy_set_header Connection 'upgrade';
+        proxy_set_header Host $host;
+        proxy_cache_bypass $http_upgrade;
+        
+        # WebSocket support
+        proxy_set_header X-Real-IP $remote_addr;
+        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
+        proxy_set_header X-Forwarded-Proto $scheme;
+    }
+    
+    # SSL configuration (recommended)
+    listen 443 ssl http2;
+    ssl_certificate /etc/letsencrypt/live/llm-proxy.yourdomain.com/fullchain.pem;
+    ssl_certificate_key /etc/letsencrypt/live/llm-proxy.yourdomain.com/privkey.pem;
+}
+```
+
+## Security Considerations
+
+### 1. Authentication
+- Use strong Bearer tokens
+- Rotate tokens regularly
+- Consider implementing JWT for production
+
+### 2. Rate Limiting
+- Implement per-client rate limiting
+- Consider using `governor` crate for advanced rate limiting
+
+### 3. Network Security
+- Run behind reverse proxy (nginx)
+- Enable HTTPS
+- Restrict access by IP if needed
+- Use firewall rules
+
+### 4. Data Security
+- Database encryption (SQLCipher for SQLite)
+- Secure API key storage
+- Regular backups
+
+## Monitoring & Maintenance
+
+### Logging
+- Application logs: `RUST_LOG=info` (or `debug` for troubleshooting)
+- Access logs via nginx
+- Database logs for audit trail
+
+### Health Checks
+```bash
+# Health endpoint
+curl http://localhost:8080/health
+
+# Database check
+sqlite3 ./data/llm_proxy.db "SELECT COUNT(*) FROM llm_requests;"
+```
+
+### Backup Strategy
+```bash
+#!/bin/bash
+# backup.sh
+BACKUP_DIR="/backups/llm-proxy"
+DATE=$(date +%Y%m%d_%H%M%S)
+
+# Backup database
+sqlite3 ./data/llm_proxy.db ".backup $BACKUP_DIR/llm_proxy_$DATE.db"
+
+# Backup configuration
+cp config.toml $BACKUP_DIR/config_$DATE.toml
+
+# Rotate old backups (keep 30 days)
+find $BACKUP_DIR -name "*.db" -mtime +30 -delete
+find $BACKUP_DIR -name "*.toml" -mtime +30 -delete
+```
+
+## Performance Tuning
+
+### Database Optimization
+```sql
+-- Run these SQL commands periodically
+VACUUM;
+ANALYZE;
+```
+
+### Memory Management
+- Monitor memory usage with `htop` or `ps aux`
+- Adjust `max_connections` based on load
+- Consider connection pooling for high traffic
+
+### Scaling
+1. **Vertical Scaling**: Increase container resources
+2. **Horizontal Scaling**: Deploy multiple instances behind load balancer
+3. **Database**: Migrate to PostgreSQL for high-volume usage
+
+## Troubleshooting
+
+### Common Issues
+
+1. **Port already in use**
+   ```bash
+   netstat -tulpn | grep :8080
+   kill <PID>  # or change port in config
+   ```
+
+2. **Database permissions**
+   ```bash
+   chown -R llmproxy:llmproxy /opt/llm-proxy/data
+   chmod 600 /opt/llm-proxy/data/llm_proxy.db
+   ```
+
+3. **API key errors**
+   - Verify environment variables are set
+   - Check provider status (dashboard)
+   - Test connectivity: `curl https://api.openai.com/v1/models`
+
+4. **High memory usage**
+   - Check for memory leaks
+   - Reduce `max_connections`
+   - Implement connection timeouts
+
+### Debug Mode
+```bash
+# Run with debug logging
+RUST_LOG=debug ./llm-proxy
+
+# Check system logs
+journalctl -u llm-proxy -f
+```
+
+## Integration
+
+### Open-WebUI Compatibility
+The proxy provides OpenAI-compatible API, so configure Open-WebUI:
+```
+API Base URL: http://your-proxy-address:8080
+API Key: sk-test-123 (or your configured token)
+```
+
+### Custom Clients
+```python
+import openai
+
+client = openai.OpenAI(
+    base_url="http://localhost:8080/v1",
+    api_key="sk-test-123"
+)
+
+response = client.chat.completions.create(
+    model="gpt-4",
+    messages=[{"role": "user", "content": "Hello"}]
+)
+```
+
+## Updates & Upgrades
+
+1. **Backup** current configuration and database
+2. **Stop** the service: `systemctl stop llm-proxy`
+3. **Update** code: `git pull` or copy new binaries
+4. **Migrate** database if needed (check migrations/)
+5. **Restart**: `systemctl start llm-proxy`
+6. **Verify**: Check logs and test endpoints
+
+## Support
+- Check logs in `/var/log/llm-proxy/`
+- Monitor dashboard at `http://your-server:8080`
+- Review database metrics in dashboard
+- Enable debug logging for troubleshooting

src/auth/mod.rs

@@ -0,0 +1,46 @@
+use axum::{extract::FromRequestParts, http::request::Parts};
+use axum_extra::headers::Authorization;
+use axum_extra::TypedHeader;
+use headers::authorization::Bearer;
+
+use crate::errors::AppError;
+
+pub struct AuthenticatedClient {
+    pub token: String,
+    pub client_id: String,
+}
+
+impl<S> FromRequestParts<S> for AuthenticatedClient
+where
+    S: Send + Sync,
+{
+    type Rejection = AppError;
+
+    fn from_request_parts(parts: &mut Parts, state: &S) -> impl std::future::Future<Output = Result<Self, Self::Rejection>> + Send {
+        async move {
+            // Extract bearer token from Authorization header
+            let TypedHeader(Authorization(bearer)) =
+                TypedHeader::<Authorization<Bearer>>::from_request_parts(parts, state)
+                    .await
+                    .map_err(|_| AppError::AuthError("Missing or invalid bearer token".to_string()))?;
+
+            let token = bearer.token().to_string();
+            
+            // In a real implementation, we would:
+            // 1. Validate token against database or config
+            // 2. Look up client_id associated with token
+            // 3. Check token permissions/rate limits
+            
+            // For now, use token hash as client_id
+            let client_id = format!("client_{}", &token[..8]);
+            
+            Ok(AuthenticatedClient { token, client_id })
+        }
+    }
+}
+
+pub fn validate_token(token: &str, valid_tokens: &[String]) -> bool {
+    // Simple validation against list of tokens
+    // In production, use proper token validation (JWT, database lookup, etc.)
+    valid_tokens.contains(&token.to_string())
+}

src/client/mod.rs

@@ -0,0 +1,310 @@
+//! Client management for LLM proxy
+//!
+//! This module handles:
+//! 1. Client registration and management
+//! 2. Client usage tracking
+//! 3. Client rate limit configuration
+
+use chrono::{DateTime, Utc};
+use serde::{Deserialize, Serialize};
+use sqlx::{SqlitePool, Row};
+use anyhow::Result;
+use tracing::{info, warn};
+
+/// Client information
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct Client {
+    pub id: i64,
+    pub client_id: String,
+    pub name: String,
+    pub description: String,
+    pub created_at: DateTime<Utc>,
+    pub updated_at: DateTime<Utc>,
+    pub is_active: bool,
+    pub rate_limit_per_minute: i64,
+    pub total_requests: i64,
+    pub total_tokens: i64,
+    pub total_cost: f64,
+}
+
+/// Client creation request
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct CreateClientRequest {
+    pub client_id: String,
+    pub name: String,
+    pub description: String,
+    pub rate_limit_per_minute: Option<i64>,
+}
+
+/// Client update request
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct UpdateClientRequest {
+    pub name: Option<String>,
+    pub description: Option<String>,
+    pub is_active: Option<bool>,
+    pub rate_limit_per_minute: Option<i64>,
+}
+
+/// Client manager for database operations
+pub struct ClientManager {
+    db_pool: SqlitePool,
+}
+
+impl ClientManager {
+    pub fn new(db_pool: SqlitePool) -> Self {
+        Self { db_pool }
+    }
+
+    /// Create a new client
+    pub async fn create_client(&self, request: CreateClientRequest) -> Result<Client> {
+        let rate_limit = request.rate_limit_per_minute.unwrap_or(60);
+        
+        // First insert the client
+        sqlx::query(
+            r#"
+            INSERT INTO clients (client_id, name, description, rate_limit_per_minute)
+            VALUES (?, ?, ?, ?)
+            "#,
+        )
+        .bind(&request.client_id)
+        .bind(&request.name)
+        .bind(&request.description)
+        .bind(rate_limit)
+        .execute(&self.db_pool)
+        .await?;
+        
+        // Then fetch the created client
+        let client = self.get_client(&request.client_id).await?
+            .ok_or_else(|| anyhow::anyhow!("Failed to retrieve created client"))?;
+        
+        info!("Created client: {} ({})", client.name, client.client_id);
+        Ok(client)
+    }
+
+    /// Get a client by ID
+    pub async fn get_client(&self, client_id: &str) -> Result<Option<Client>> {
+        let row = sqlx::query(
+            r#"
+            SELECT 
+                id, client_id, name, description,
+                created_at, updated_at, is_active,
+                rate_limit_per_minute, total_requests, total_tokens, total_cost
+            FROM clients
+            WHERE client_id = ?
+            "#,
+        )
+        .bind(client_id)
+        .fetch_optional(&self.db_pool)
+        .await?;
+        
+        if let Some(row) = row {
+            let client = Client {
+                id: row.get("id"),
+                client_id: row.get("client_id"),
+                name: row.get("name"),
+                description: row.get("description"),
+                created_at: row.get("created_at"),
+                updated_at: row.get("updated_at"),
+                is_active: row.get("is_active"),
+                rate_limit_per_minute: row.get("rate_limit_per_minute"),
+                total_requests: row.get("total_requests"),
+                total_tokens: row.get("total_tokens"),
+                total_cost: row.get("total_cost"),
+            };
+            Ok(Some(client))
+        } else {
+            Ok(None)
+        }
+    }
+
+    /// Update a client
+    pub async fn update_client(&self, client_id: &str, request: UpdateClientRequest) -> Result<Option<Client>> {
+        // First, get the current client to check if it exists
+        let current_client = self.get_client(client_id).await?;
+        if current_client.is_none() {
+            return Ok(None);
+        }
+        
+        // Build update query dynamically based on provided fields
+        let mut updates = Vec::new();
+        let mut query_builder = sqlx::QueryBuilder::new("UPDATE clients SET ");
+        let mut has_updates = false;
+        
+        if let Some(name) = &request.name {
+            updates.push("name = ");
+            query_builder.push_bind(name);
+            has_updates = true;
+        }
+        
+        if let Some(description) = &request.description {
+            if has_updates {
+                query_builder.push(", ");
+            }
+            updates.push("description = ");
+            query_builder.push_bind(description);
+            has_updates = true;
+        }
+        
+        if let Some(is_active) = request.is_active {
+            if has_updates {
+                query_builder.push(", ");
+            }
+            updates.push("is_active = ");
+            query_builder.push_bind(is_active);
+            has_updates = true;
+        }
+        
+        if let Some(rate_limit) = request.rate_limit_per_minute {
+            if has_updates {
+                query_builder.push(", ");
+            }
+            updates.push("rate_limit_per_minute = ");
+            query_builder.push_bind(rate_limit);
+            has_updates = true;
+        }
+        
+        // Always update the updated_at timestamp
+        if has_updates {
+            query_builder.push(", ");
+        }
+        query_builder.push("updated_at = CURRENT_TIMESTAMP");
+        
+        if !has_updates {
+            // No updates to make
+            return self.get_client(client_id).await;
+        }
+        
+        query_builder.push(" WHERE client_id = ");
+        query_builder.push_bind(client_id);
+        
+        let query = query_builder.build();
+        query.execute(&self.db_pool).await?;
+        
+        // Fetch the updated client
+        let updated_client = self.get_client(client_id).await?;
+        
+        if updated_client.is_some() {
+            info!("Updated client: {}", client_id);
+        }
+        
+        Ok(updated_client)
+    }
+
+    /// List all clients
+    pub async fn list_clients(&self, limit: Option<i64>, offset: Option<i64>) -> Result<Vec<Client>> {
+        let limit = limit.unwrap_or(100);
+        let offset = offset.unwrap_or(0);
+        
+        let rows = sqlx::query(
+            r#"
+            SELECT 
+                id, client_id, name, description,
+                created_at, updated_at, is_active,
+                rate_limit_per_minute, total_requests, total_tokens, total_cost
+            FROM clients
+            ORDER BY created_at DESC
+            LIMIT ? OFFSET ?
+            "#
+        )
+        .bind(limit)
+        .bind(offset)
+        .fetch_all(&self.db_pool)
+        .await?;
+        
+        let mut clients = Vec::new();
+        for row in rows {
+            let client = Client {
+                id: row.get("id"),
+                client_id: row.get("client_id"),
+                name: row.get("name"),
+                description: row.get("description"),
+                created_at: row.get("created_at"),
+                updated_at: row.get("updated_at"),
+                is_active: row.get("is_active"),
+                rate_limit_per_minute: row.get("rate_limit_per_minute"),
+                total_requests: row.get("total_requests"),
+                total_tokens: row.get("total_tokens"),
+                total_cost: row.get("total_cost"),
+            };
+            clients.push(client);
+        }
+        
+        Ok(clients)
+    }
+
+    /// Delete a client
+    pub async fn delete_client(&self, client_id: &str) -> Result<bool> {
+        let result = sqlx::query(
+            "DELETE FROM clients WHERE client_id = ?"
+        )
+        .bind(client_id)
+        .execute(&self.db_pool)
+        .await?;
+        
+        let deleted = result.rows_affected() > 0;
+        
+        if deleted {
+            info!("Deleted client: {}", client_id);
+        } else {
+            warn!("Client not found for deletion: {}", client_id);
+        }
+        
+        Ok(deleted)
+    }
+
+    /// Update client usage statistics after a request
+    pub async fn update_client_usage(
+        &self,
+        client_id: &str,
+        tokens: i64,
+        cost: f64,
+    ) -> Result<()> {
+        sqlx::query(
+            r#"
+            UPDATE clients 
+            SET 
+                total_requests = total_requests + 1,
+                total_tokens = total_tokens + ?,
+                total_cost = total_cost + ?,
+                updated_at = CURRENT_TIMESTAMP
+            WHERE client_id = ?
+            "#
+        )
+        .bind(tokens)
+        .bind(cost)
+        .bind(client_id)
+        .execute(&self.db_pool)
+        .await?;
+        
+        Ok(())
+    }
+
+    /// Get client usage statistics
+    pub async fn get_client_usage(&self, client_id: &str) -> Result<Option<(i64, i64, f64)>> {
+        let row = sqlx::query(
+            r#"
+            SELECT total_requests, total_tokens, total_cost
+            FROM clients
+            WHERE client_id = ?
+            "#
+        )
+        .bind(client_id)
+        .fetch_optional(&self.db_pool)
+        .await?;
+        
+        if let Some(row) = row {
+            let total_requests: i64 = row.get("total_requests");
+            let total_tokens: i64 = row.get("total_tokens");
+            let total_cost: f64 = row.get("total_cost");
+            Ok(Some((total_requests, total_tokens, total_cost)))
+        } else {
+            Ok(None)
+        }
+    }
+
+    /// Check if a client exists and is active
+    pub async fn validate_client(&self, client_id: &str) -> Result<bool> {
+        let client = self.get_client(client_id).await?;
+        Ok(client.map(|c| c.is_active).unwrap_or(false))
+    }
+}

src/config/mod.rs

@@ -0,0 +1,191 @@
+use anyhow::Result;
+use config::{Config, File, FileFormat};
+use serde::{Deserialize, Serialize};
+use std::path::PathBuf;
+use std::sync::Arc;
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ServerConfig {
+    pub port: u16,
+    pub host: String,
+    pub auth_tokens: Vec<String>,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct DatabaseConfig {
+    pub path: PathBuf,
+    pub max_connections: u32,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ProviderConfig {
+    pub openai: OpenAIConfig,
+    pub gemini: GeminiConfig,
+    pub deepseek: DeepSeekConfig,
+    pub grok: GrokConfig,
+    pub ollama: OllamaConfig,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct OpenAIConfig {
+    pub api_key_env: String,
+    pub base_url: String,
+    pub default_model: String,
+    pub enabled: bool,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct GeminiConfig {
+    pub api_key_env: String,
+    pub base_url: String,
+    pub default_model: String,
+    pub enabled: bool,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct DeepSeekConfig {
+    pub api_key_env: String,
+    pub base_url: String,
+    pub default_model: String,
+    pub enabled: bool,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct GrokConfig {
+    pub api_key_env: String,
+    pub base_url: String,
+    pub default_model: String,
+    pub enabled: bool,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct OllamaConfig {
+    pub base_url: String,
+    pub enabled: bool,
+    pub models: Vec<String>,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ModelMappingConfig {
+    pub patterns: Vec<(String, String)>,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct PricingConfig {
+    pub openai: Vec<ModelPricing>,
+    pub gemini: Vec<ModelPricing>,
+    pub deepseek: Vec<ModelPricing>,
+    pub grok: Vec<ModelPricing>,
+    pub ollama: Vec<ModelPricing>,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ModelPricing {
+    pub model: String,
+    pub prompt_tokens_per_million: f64,
+    pub completion_tokens_per_million: f64,
+}
+
+#[derive(Debug, Clone)]
+pub struct AppConfig {
+    pub server: ServerConfig,
+    pub database: DatabaseConfig,
+    pub providers: ProviderConfig,
+    pub model_mapping: ModelMappingConfig,
+    pub pricing: PricingConfig,
+    pub config_path: PathBuf,
+}
+
+impl AppConfig {
+    pub async fn load() -> Result<Arc<Self>> {
+        Self::load_from_path(None).await
+    }
+    
+    /// Load configuration from a specific path (for testing)
+    pub async fn load_from_path(config_path: Option<PathBuf>) -> Result<Arc<Self>> {
+        // Load configuration from multiple sources
+        let mut config_builder = Config::builder();
+
+        // Default configuration
+        config_builder = config_builder
+            .set_default("server.port", 8080)?
+            .set_default("server.host", "0.0.0.0")?
+            .set_default("server.auth_tokens", Vec::<String>::new())?
+            .set_default("database.path", "./data/llm_proxy.db")?
+            .set_default("database.max_connections", 10)?
+            .set_default("providers.openai.api_key_env", "OPENAI_API_KEY")?
+            .set_default("providers.openai.base_url", "https://api.openai.com/v1")?
+            .set_default("providers.openai.default_model", "gpt-4o")?
+            .set_default("providers.openai.enabled", true)?
+            .set_default("providers.gemini.api_key_env", "GEMINI_API_KEY")?
+            .set_default("providers.gemini.base_url", "https://generativelanguage.googleapis.com/v1")?
+            .set_default("providers.gemini.default_model", "gemini-2.0-flash")?
+            .set_default("providers.gemini.enabled", true)?
+            .set_default("providers.deepseek.api_key_env", "DEEPSEEK_API_KEY")?
+            .set_default("providers.deepseek.base_url", "https://api.deepseek.com")?
+            .set_default("providers.deepseek.default_model", "deepseek-reasoner")?
+            .set_default("providers.deepseek.enabled", true)?
+            .set_default("providers.grok.api_key_env", "GROK_API_KEY")?
+            .set_default("providers.grok.base_url", "https://api.x.ai/v1")?
+            .set_default("providers.grok.default_model", "grok-beta")?
+            .set_default("providers.grok.enabled", false)?
+            .set_default("providers.ollama.base_url", "http://localhost:11434/v1")?
+            .set_default("providers.ollama.enabled", false)?
+            .set_default("providers.ollama.models", Vec::<String>::new())?;
+
+        // Load from config file if exists
+        let config_path = config_path.unwrap_or_else(|| std::env::current_dir().unwrap().join("config.toml"));
+        if config_path.exists() {
+            config_builder = config_builder.add_source(File::from(config_path.clone()).format(FileFormat::Toml));
+        }
+
+        // Load from .env file
+        dotenvy::dotenv().ok();
+
+        // Load from environment variables (with prefix "LLM_PROXY_")
+        config_builder = config_builder.add_source(
+            config::Environment::with_prefix("LLM_PROXY")
+                .separator("__")
+                .try_parsing(true),
+        );
+
+        let config = config_builder.build()?;
+
+        // Deserialize configuration
+        let server: ServerConfig = config.get("server")?;
+        let database: DatabaseConfig = config.get("database")?;
+        let providers: ProviderConfig = config.get("providers")?;
+        
+        // For now, use empty model mapping and pricing (will be populated later)
+        let model_mapping = ModelMappingConfig { patterns: vec![] };
+        let pricing = PricingConfig {
+            openai: vec![],
+            gemini: vec![],
+            deepseek: vec![],
+            grok: vec![],
+            ollama: vec![],
+        };
+
+        Ok(Arc::new(AppConfig {
+            server,
+            database,
+            providers,
+            model_mapping,
+            pricing,
+            config_path,
+        }))
+    }
+
+    pub fn get_api_key(&self, provider: &str) -> Result<String> {
+        let env_var = match provider {
+            "openai" => &self.providers.openai.api_key_env,
+            "gemini" => &self.providers.gemini.api_key_env,
+            "deepseek" => &self.providers.deepseek.api_key_env,
+            "grok" => &self.providers.grok.api_key_env,
+            _ => return Err(anyhow::anyhow!("Unknown provider: {}", provider)),
+        };
+
+        std::env::var(env_var)
+            .map_err(|_| anyhow::anyhow!("Environment variable {} not set for {}", env_var, provider))
+    }
+}

src/dashboard/mod.rs

@@ -0,0 +1,642 @@
+// Dashboard module for LLM Proxy Gateway
+
+use axum::{
+    extract::{ws::{Message, WebSocket, WebSocketUpgrade}, State},
+    response::{IntoResponse, Json},
+    routing::{get, post},
+    Router,
+};
+use serde::Serialize;
+use sqlx::Row;
+use std::collections::HashMap;
+use tracing::{info, warn};
+
+use crate::state::AppState;
+
+// Dashboard state
+#[derive(Clone)]
+struct DashboardState {
+    app_state: AppState,
+}
+
+// API Response types
+#[derive(Serialize)]
+struct ApiResponse<T> {
+    success: bool,
+    data: Option<T>,
+    error: Option<String>,
+}
+
+impl<T> ApiResponse<T> {
+    fn success(data: T) -> Self {
+        Self {
+            success: true,
+            data: Some(data),
+            error: None,
+        }
+    }
+    
+    fn error(error: String) -> Self {
+        Self {
+            success: false,
+            data: None,
+            error: Some(error),
+        }
+    }
+}
+
+// ... (keep routes as they are)
+
+// Dashboard routes
+pub fn router(state: AppState) -> Router {
+    let dashboard_state = DashboardState {
+        app_state: state,
+    };
+    
+    Router::new()
+        // Static file serving
+        .nest_service("/", tower_http::services::ServeDir::new("static"))
+        .fallback_service(tower_http::services::ServeDir::new("static"))
+        
+        // WebSocket endpoint
+        .route("/ws", get(handle_websocket))
+        
+        // API endpoints
+        .route("/api/auth/login", post(handle_login))
+        .route("/api/auth/status", get(handle_auth_status))
+        .route("/api/usage/summary", get(handle_usage_summary))
+        .route("/api/usage/time-series", get(handle_time_series))
+        .route("/api/usage/clients", get(handle_clients_usage))
+        .route("/api/usage/providers", get(handle_providers_usage))
+        .route("/api/clients", get(handle_get_clients).post(handle_create_client))
+        .route("/api/clients/:id", get(handle_get_client).delete(handle_delete_client))
+        .route("/api/clients/:id/usage", get(handle_client_usage))
+        .route("/api/providers", get(handle_get_providers))
+        .route("/api/providers/:name", get(handle_get_provider).put(handle_update_provider))
+        .route("/api/providers/:name/test", post(handle_test_provider))
+        .route("/api/system/health", get(handle_system_health))
+        .route("/api/system/logs", get(handle_system_logs))
+        .route("/api/system/backup", post(handle_system_backup))
+        
+        .with_state(dashboard_state)
+}
+
+// WebSocket handler
+async fn handle_websocket(
+    ws: WebSocketUpgrade,
+    State(state): State<DashboardState>,
+) -> impl IntoResponse {
+    ws.on_upgrade(|socket| handle_websocket_connection(socket, state))
+}
+
+async fn handle_websocket_connection(mut socket: WebSocket, state: DashboardState) {
+    info!("WebSocket connection established");
+    
+    // Subscribe to events from the global bus
+    let mut rx = state.app_state.dashboard_tx.subscribe();
+    
+    // Send initial connection message
+    let _ = socket.send(Message::Text(
+        serde_json::json!({
+            "type": "connected",
+            "message": "Connected to LLM Proxy Dashboard"
+        }).to_string().into(),
+    )).await;
+    
+    // Handle incoming messages and broadcast events
+    loop {
+        tokio::select! {
+            // Receive broadcast events
+            Ok(event) = rx.recv() => {
+                let message = Message::Text(serde_json::to_string(&event).unwrap().into());
+                if socket.send(message).await.is_err() {
+                    break;
+                }
+            }
+            
+            // Receive WebSocket messages
+            result = socket.recv() => {
+                match result {
+                    Some(Ok(Message::Text(text))) => {
+                        handle_websocket_message(&text, &state).await;
+                    }
+                    _ => break,
+                }
+            }
+        }
+    }
+    
+    info!("WebSocket connection closed");
+}
+
+async fn handle_websocket_message(text: &str, state: &DashboardState) {
+    // Parse and handle WebSocket messages
+    if let Ok(data) = serde_json::from_str::<serde_json::Value>(text) {
+        if let Some("ping") = data.get("type").and_then(|v| v.as_str()) {
+            let _ = state.app_state.dashboard_tx.send(serde_json::json!({
+                "event_type": "pong",
+                "data": {}
+            }));
+        }
+    }
+}
+
+// Authentication handlers
+async fn handle_login() -> Json<ApiResponse<serde_json::Value>> {
+    // Simple authentication for demo
+    // In production, this would validate credentials against a database
+    Json(ApiResponse::success(serde_json::json!({
+        "token": "demo-token-123456",
+        "user": {
+            "username": "admin",
+            "name": "Administrator",
+            "role": "Super Admin"
+        }
+    })))
+}
+
+async fn handle_auth_status() -> Json<ApiResponse<serde_json::Value>> {
+    Json(ApiResponse::success(serde_json::json!({
+        "authenticated": true,
+        "user": {
+            "username": "admin",
+            "name": "Administrator",
+            "role": "Super Admin"
+        }
+    })))
+}
+
+// Usage handlers
+async fn handle_usage_summary(State(state): State<DashboardState>) -> Json<ApiResponse<serde_json::Value>> {
+    let pool = &state.app_state.db_pool;
+    
+    // Total stats
+    let total_stats = sqlx::query(
+        r#"
+        SELECT 
+            COUNT(*) as total_requests,
+            COALESCE(SUM(total_tokens), 0) as total_tokens,
+            COALESCE(SUM(cost), 0.0) as total_cost,
+            COUNT(DISTINCT client_id) as active_clients
+        FROM llm_requests
+        "#
+    )
+    .fetch_one(pool);
+    
+    // Today's stats
+    let today = chrono::Utc::now().format("%Y-%m-%d").to_string();
+    let today_stats = sqlx::query(
+        r#"
+        SELECT 
+            COUNT(*) as today_requests,
+            COALESCE(SUM(total_tokens), 0) as today_tokens,
+            COALESCE(SUM(cost), 0.0) as today_cost
+        FROM llm_requests
+        WHERE strftime('%Y-%m-%d', timestamp) = ?
+        "#
+    )
+    .bind(today)
+    .fetch_one(pool);
+    
+    // Error stats
+    let error_stats = sqlx::query(
+        r#"
+        SELECT 
+            COUNT(*) as total,
+            SUM(CASE WHEN status = 'error' THEN 1 ELSE 0 END) as errors
+        FROM llm_requests
+        "#
+    )
+    .fetch_one(pool);
+    
+    // Average response time
+    let avg_response = sqlx::query(
+        r#"
+        SELECT COALESCE(AVG(duration_ms), 0.0) as avg_duration
+        FROM llm_requests
+        WHERE status = 'success'
+        "#
+    )
+    .fetch_one(pool);
+    
+    match tokio::join!(total_stats, today_stats, error_stats, avg_response) {
+        (Ok(t), Ok(d), Ok(e), Ok(a)) => {
+            let total_requests: i64 = t.get("total_requests");
+            let total_tokens: i64 = t.get("total_tokens");
+            let total_cost: f64 = t.get("total_cost");
+            let active_clients: i64 = t.get("active_clients");
+            
+            let today_requests: i64 = d.get("today_requests");
+            let today_cost: f64 = d.get("today_cost");
+            
+            let total_count: i64 = e.get("total");
+            let error_count: i64 = e.get("errors");
+            let error_rate = if total_count > 0 {
+                (error_count as f64 / total_count as f64) * 100.0
+            } else {
+                0.0
+            };
+            
+            let avg_response_time: f64 = a.get("avg_duration");
+            
+            Json(ApiResponse::success(serde_json::json!({
+                "total_requests": total_requests,
+                "total_tokens": total_tokens,
+                "total_cost": total_cost,
+                "active_clients": active_clients,
+                "today_requests": today_requests,
+                "today_cost": today_cost,
+                "error_rate": error_rate,
+                "avg_response_time": avg_response_time,
+            })))
+        }
+        _ => Json(ApiResponse::error("Failed to fetch usage statistics".to_string()))
+    }
+}
+
+async fn handle_time_series(State(state): State<DashboardState>) -> Json<ApiResponse<serde_json::Value>> {
+    let pool = &state.app_state.db_pool;
+    
+    let now = chrono::Utc::now();
+    let twenty_four_hours_ago = now - chrono::Duration::hours(24);
+    
+    let result = sqlx::query(
+        r#"
+        SELECT 
+            strftime('%H:00', timestamp) as hour,
+            COUNT(*) as requests,
+            SUM(total_tokens) as tokens,
+            SUM(cost) as cost
+        FROM llm_requests
+        WHERE timestamp >= ?
+        GROUP BY hour
+        ORDER BY hour
+        "#
+    )
+    .bind(twenty_four_hours_ago)
+    .fetch_all(pool)
+    .await;
+    
+    match result {
+        Ok(rows) => {
+            let mut series = Vec::new();
+            
+            for row in rows {
+                let hour: String = row.get("hour");
+                let requests: i64 = row.get("requests");
+                let tokens: i64 = row.get("tokens");
+                let cost: f64 = row.get("cost");
+                
+                series.push(serde_json::json!({
+                    "time": hour,
+                    "requests": requests,
+                    "tokens": tokens,
+                    "cost": cost,
+                }));
+            }
+            
+            Json(ApiResponse::success(serde_json::json!({
+                "series": series,
+                "period": "24h"
+            })))
+        }
+        Err(e) => {
+            warn!("Failed to fetch time series data: {}", e);
+            Json(ApiResponse::error("Failed to fetch time series data".to_string()))
+        }
+    }
+}
+
+async fn handle_clients_usage(State(state): State<DashboardState>) -> Json<ApiResponse<serde_json::Value>> {
+    // Query database for client usage statistics
+    let pool = &state.app_state.db_pool;
+    
+    let result = sqlx::query(
+        r#"
+        SELECT 
+            client_id,
+            COUNT(*) as requests,
+            SUM(total_tokens) as tokens,
+            SUM(cost) as cost,
+            MAX(timestamp) as last_request
+        FROM llm_requests
+        GROUP BY client_id
+        ORDER BY requests DESC
+        "#
+    )
+    .fetch_all(pool)
+    .await;
+    
+    match result {
+        Ok(rows) => {
+            let mut client_usage = Vec::new();
+            
+            for row in rows {
+                let client_id: String = row.get("client_id");
+                let requests: i64 = row.get("requests");
+                let tokens: i64 = row.get("tokens");
+                let cost: f64 = row.get("cost");
+                let last_request: Option<chrono::DateTime<chrono::Utc>> = row.get("last_request");
+                
+                client_usage.push(serde_json::json!({
+                    "client_id": client_id,
+                    "client_name": client_id,
+                    "requests": requests,
+                    "tokens": tokens,
+                    "cost": cost,
+                    "last_request": last_request,
+                }));
+            }
+            
+            Json(ApiResponse::success(serde_json::json!(client_usage)))
+        }
+        Err(e) => {
+            warn!("Failed to fetch client usage data: {}", e);
+            Json(ApiResponse::error("Failed to fetch client usage data".to_string()))
+        }
+    }
+}
+
+async fn handle_providers_usage(State(state): State<DashboardState>) -> Json<ApiResponse<serde_json::Value>> {
+    // Query database for provider usage statistics
+    let pool = &state.app_state.db_pool;
+    
+    let result = sqlx::query(
+        r#"
+        SELECT 
+            provider,
+            COUNT(*) as requests,
+            COALESCE(SUM(total_tokens), 0) as tokens,
+            COALESCE(SUM(cost), 0.0) as cost
+        FROM llm_requests
+        GROUP BY provider
+        ORDER BY requests DESC
+        "#
+    )
+    .fetch_all(pool)
+    .await;
+    
+    match result {
+        Ok(rows) => {
+            let mut provider_usage = Vec::new();
+            
+            for row in rows {
+                let provider: String = row.get("provider");
+                let requests: i64 = row.get("requests");
+                let tokens: i64 = row.get("tokens");
+                let cost: f64 = row.get("cost");
+                
+                provider_usage.push(serde_json::json!({
+                    "provider": provider,
+                    "requests": requests,
+                    "tokens": tokens,
+                    "cost": cost,
+                }));
+            }
+            
+            Json(ApiResponse::success(serde_json::json!(provider_usage)))
+        }
+        Err(e) => {
+            warn!("Failed to fetch provider usage data: {}", e);
+            Json(ApiResponse::error("Failed to fetch provider usage data".to_string()))
+        }
+    }
+}
+
+// Client handlers
+async fn handle_get_clients(State(state): State<DashboardState>) -> Json<ApiResponse<serde_json::Value>> {
+    let pool = &state.app_state.db_pool;
+    
+    let result = sqlx::query(
+        r#"
+        SELECT 
+            client_id as id,
+            name,
+            created_at,
+            total_requests,
+            total_tokens,
+            total_cost,
+            is_active
+        FROM clients
+        ORDER BY created_at DESC
+        "#
+    )
+    .fetch_all(pool)
+    .await;
+    
+    match result {
+        Ok(rows) => {
+            let clients: Vec<serde_json::Value> = rows.into_iter().map(|row| {
+                serde_json::json!({
+                    "id": row.get::<String, _>("id"),
+                    "name": row.get::<Option<String>, _>("name").unwrap_or_else(|| "Unnamed".to_string()),
+                    "created_at": row.get::<chrono::DateTime<chrono::Utc>, _>("created_at"),
+                    "requests_count": row.get::<i64, _>("total_requests"),
+                    "total_tokens": row.get::<i64, _>("total_tokens"),
+                    "total_cost": row.get::<f64, _>("total_cost"),
+                    "status": if row.get::<bool, _>("is_active") { "active" } else { "inactive" },
+                })
+            }).collect();
+            
+            Json(ApiResponse::success(serde_json::json!(clients)))
+        }
+        Err(e) => {
+            warn!("Failed to fetch clients: {}", e);
+            Json(ApiResponse::error("Failed to fetch clients".to_string()))
+        }
+    }
+}
+
+async fn handle_create_client() -> Json<ApiResponse<serde_json::Value>> {
+    // In production, this would create a real client
+    Json(ApiResponse::success(serde_json::json!({
+        "id": format!("client-{}", rand::random::<u32>()),
+        "name": "New Client",
+        "token": format!("sk-demo-{}", rand::random::<u32>()),
+        "created_at": chrono::Utc::now().to_rfc3339(),
+        "last_used": None::<String>,
+        "requests_count": 0,
+        "status": "active",
+    })))
+}
+
+async fn handle_get_client() -> Json<ApiResponse<serde_json::Value>> {
+    Json(ApiResponse::error("Not implemented".to_string()))
+}
+
+async fn handle_delete_client() -> Json<ApiResponse<serde_json::Value>> {
+    Json(ApiResponse::success(serde_json::json!({
+        "success": true,
+        "message": "Client deleted"
+    })))
+}
+
+async fn handle_client_usage() -> Json<ApiResponse<serde_json::Value>> {
+    Json(ApiResponse::error("Not implemented".to_string()))
+}
+
+// Provider handlers
+async fn handle_get_providers(State(state): State<DashboardState>) -> Json<ApiResponse<serde_json::Value>> {
+    let registry = &state.app_state.model_registry;
+    
+    let mut providers_json = Vec::new();
+    
+    for (p_id, p_info) in &registry.providers {
+        let models: Vec<String> = p_info.models.keys().cloned().collect();
+        
+        // Check if provider is healthy via circuit breaker
+        let status = if state.app_state.rate_limit_manager.check_provider_request(p_id).await.unwrap_or(true) {
+            "online"
+        } else {
+            "degraded"
+        };
+
+        providers_json.push(serde_json::json!({
+            "id": p_id,
+            "name": p_info.name,
+            "enabled": true,
+            "status": status,
+            "models": models,
+            "last_used": null, // TODO: track last used
+        }));
+    }
+
+    // Add Ollama explicitly
+    providers_json.push(serde_json::json!({
+        "id": "ollama",
+        "name": "Ollama",
+        "enabled": true,
+        "status": "online",
+        "models": ["llama3", "mistral", "phi3"], 
+        "last_used": null,
+    }));
+    
+    Json(ApiResponse::success(serde_json::json!(providers_json)))
+}
+
+async fn handle_get_provider() -> Json<ApiResponse<serde_json::Value>> {
+    Json(ApiResponse::error("Not implemented".to_string()))
+}
+
+async fn handle_update_provider() -> Json<ApiResponse<serde_json::Value>> {
+    Json(ApiResponse::success(serde_json::json!({
+        "success": true,
+        "message": "Provider updated"
+    })))
+}
+
+async fn handle_test_provider() -> Json<ApiResponse<serde_json::Value>> {
+    Json(ApiResponse::success(serde_json::json!({
+        "success": true,
+        "latency": rand::random::<u32>() % 500 + 100,
+        "message": "Connection test successful"
+    })))
+}
+
+// System handlers
+async fn handle_system_health(State(state): State<DashboardState>) -> Json<ApiResponse<serde_json::Value>> {
+    let mut components = HashMap::new();
+    components.insert("api_server", "online");
+    components.insert("database", "online");
+    
+    // Check provider health via circuit breakers
+    for p_id in state.app_state.model_registry.providers.keys() {
+        if state.app_state.rate_limit_manager.check_provider_request(p_id).await.unwrap_or(true) {
+            components.insert(p_id.as_str(), "online");
+        } else {
+            components.insert(p_id.as_str(), "degraded");
+        }
+    }
+
+    // Check Ollama health
+    if state.app_state.rate_limit_manager.check_provider_request("ollama").await.unwrap_or(true) {
+        components.insert("ollama", "online");
+    } else {
+        components.insert("ollama", "degraded");
+    }
+
+    Json(ApiResponse::success(serde_json::json!({
+        "status": "healthy",
+        "timestamp": chrono::Utc::now().to_rfc3339(),
+        "components": components,
+        "metrics": {
+            "cpu_usage": rand::random::<f64>() * 10.0 + 5.0,
+            "memory_usage": rand::random::<f64>() * 20.0 + 40.0,
+            "active_connections": rand::random::<u32>() % 20 + 5,
+        }
+    })))
+}
+
+async fn handle_system_logs(State(state): State<DashboardState>) -> Json<ApiResponse<serde_json::Value>> {
+    let pool = &state.app_state.db_pool;
+    
+    let result = sqlx::query(
+        r#"
+        SELECT 
+            id,
+            timestamp,
+            client_id,
+            provider,
+            model,
+            prompt_tokens,
+            completion_tokens,
+            total_tokens,
+            cost,
+            status,
+            error_message,
+            duration_ms
+        FROM llm_requests
+        ORDER BY timestamp DESC
+        LIMIT 100
+        "#
+    )
+    .fetch_all(pool)
+    .await;
+    
+    match result {
+        Ok(rows) => {
+            let logs: Vec<serde_json::Value> = rows.into_iter().map(|row| {
+                serde_json::json!({
+                    "id": row.get::<i64, _>("id"),
+                    "timestamp": row.get::<chrono::DateTime<chrono::Utc>, _>("timestamp"),
+                    "client_id": row.get::<String, _>("client_id"),
+                    "provider": row.get::<String, _>("provider"),
+                    "model": row.get::<String, _>("model"),
+                    "tokens": row.get::<i64, _>("total_tokens"),
+                    "cost": row.get::<f64, _>("cost"),
+                    "status": row.get::<String, _>("status"),
+                    "error": row.get::<Option<String>, _>("error_message"),
+                    "duration": row.get::<i64, _>("duration_ms"),
+                })
+            }).collect();
+            
+            Json(ApiResponse::success(serde_json::json!(logs)))
+        }
+        Err(e) => {
+            warn!("Failed to fetch system logs: {}", e);
+            Json(ApiResponse::error("Failed to fetch system logs".to_string()))
+        }
+    }
+}
+
+async fn handle_system_backup() -> Json<ApiResponse<serde_json::Value>> {
+    Json(ApiResponse::success(serde_json::json!({
+        "success": true,
+        "message": "Backup initiated",
+        "backup_id": format!("backup-{}", chrono::Utc::now().timestamp()),
+    })))
+}
+
+// Helper functions
+#[allow(dead_code)]
+fn mask_token(token: &str) -> String {
+    if token.len() <= 8 {
+        return "*****".to_string();
+    }
+    
+    let masked_len = token.len().min(12);
+    let visible_len = 4;
+    let mask_len = masked_len - visible_len;
+    
+    format!("{}{}", "*".repeat(mask_len), &token[token.len() - visible_len..])
+}

src/database/mod.rs

@@ -0,0 +1,128 @@
+use anyhow::Result;
+use sqlx::SqlitePool;
+use tracing::info;
+
+use crate::config::DatabaseConfig;
+
+pub type DbPool = SqlitePool;
+
+pub async fn init(config: &DatabaseConfig) -> Result<DbPool> {
+    // Ensure the database directory exists
+    if let Some(parent) = config.path.parent() {
+        tokio::fs::create_dir_all(parent).await?;
+    }
+
+    let database_url = format!("sqlite:{}", config.path.display());
+    info!("Connecting to database at {}", database_url);
+
+    let pool = SqlitePool::connect(&database_url).await?;
+
+    // Run migrations
+    run_migrations(&pool).await?;
+    info!("Database migrations completed");
+
+    Ok(pool)
+}
+
+async fn run_migrations(pool: &DbPool) -> Result<()> {
+    // Create clients table if it doesn't exist
+    sqlx::query(
+        r#"
+        CREATE TABLE IF NOT EXISTS clients (
+            id INTEGER PRIMARY KEY AUTOINCREMENT,
+            client_id TEXT UNIQUE NOT NULL,
+            name TEXT,
+            description TEXT,
+            created_at DATETIME DEFAULT CURRENT_TIMESTAMP,
+            updated_at DATETIME DEFAULT CURRENT_TIMESTAMP,
+            is_active BOOLEAN DEFAULT TRUE,
+            rate_limit_per_minute INTEGER DEFAULT 60,
+            total_requests INTEGER DEFAULT 0,
+            total_tokens INTEGER DEFAULT 0,
+            total_cost REAL DEFAULT 0.0
+        )
+        "#,
+    )
+    .execute(pool)
+    .await?;
+
+    // Create llm_requests table if it doesn't exist
+    sqlx::query(
+        r#"
+        CREATE TABLE IF NOT EXISTS llm_requests (
+            id INTEGER PRIMARY KEY AUTOINCREMENT,
+            timestamp DATETIME DEFAULT CURRENT_TIMESTAMP,
+            client_id TEXT,
+            provider TEXT,
+            model TEXT,
+            prompt_tokens INTEGER,
+            completion_tokens INTEGER,
+            total_tokens INTEGER,
+            cost REAL,
+            has_images BOOLEAN DEFAULT FALSE,
+            status TEXT DEFAULT 'success',
+            error_message TEXT,
+            duration_ms INTEGER,
+            request_body TEXT,
+            response_body TEXT,
+            FOREIGN KEY (client_id) REFERENCES clients(client_id) ON DELETE SET NULL
+        )
+        "#,
+    )
+    .execute(pool)
+    .await?;
+
+    // Create indices
+    sqlx::query(
+        "CREATE INDEX IF NOT EXISTS idx_clients_client_id ON clients(client_id)"
+    )
+    .execute(pool)
+    .await?;
+
+    sqlx::query(
+        "CREATE INDEX IF NOT EXISTS idx_clients_created_at ON clients(created_at)"
+    )
+    .execute(pool)
+    .await?;
+
+    sqlx::query(
+        "CREATE INDEX IF NOT EXISTS idx_llm_requests_timestamp ON llm_requests(timestamp)"
+    )
+    .execute(pool)
+    .await?;
+
+    sqlx::query(
+        "CREATE INDEX IF NOT EXISTS idx_llm_requests_client_id ON llm_requests(client_id)"
+    )
+    .execute(pool)
+    .await?;
+
+    sqlx::query(
+        "CREATE INDEX IF NOT EXISTS idx_llm_requests_provider ON llm_requests(provider)"
+    )
+    .execute(pool)
+    .await?;
+
+    sqlx::query(
+        "CREATE INDEX IF NOT EXISTS idx_llm_requests_status ON llm_requests(status)"
+    )
+    .execute(pool)
+    .await?;
+
+    // Insert default client if none exists
+    sqlx::query(
+        r#"
+        INSERT OR IGNORE INTO clients (client_id, name, description)
+        VALUES ('default', 'Default Client', 'Default client for anonymous requests')
+        "#,
+    )
+    .execute(pool)
+    .await?;
+
+    Ok(())
+}
+
+pub async fn test_connection(pool: &DbPool) -> Result<()> {
+    sqlx::query("SELECT 1").execute(pool).await?;
+    Ok(())
+}

src/errors/mod.rs

@@ -0,0 +1,58 @@
+use thiserror::Error;
+
+#[derive(Error, Debug, Clone)]
+pub enum AppError {
+    #[error("Authentication failed: {0}")]
+    AuthError(String),
+
+    #[error("Configuration error: {0}")]
+    ConfigError(String),
+
+    #[error("Database error: {0}")]
+    DatabaseError(String),
+
+    #[error("Provider error: {0}")]
+    ProviderError(String),
+
+    #[error("Validation error: {0}")]
+    ValidationError(String),
+
+    #[error("Multimodal processing error: {0}")]
+    MultimodalError(String),
+
+    #[error("Rate limit exceeded: {0}")]
+    RateLimitError(String),
+
+    #[error("Internal server error: {0}")]
+    InternalError(String),
+}
+
+impl From<sqlx::Error> for AppError {
+    fn from(err: sqlx::Error) -> Self {
+        AppError::DatabaseError(err.to_string())
+    }
+}
+
+impl From<anyhow::Error> for AppError {
+    fn from(err: anyhow::Error) -> Self {
+        AppError::InternalError(err.to_string())
+    }
+}
+
+impl axum::response::IntoResponse for AppError {
+    fn into_response(self) -> axum::response::Response {
+        let status = match self {
+            AppError::AuthError(_) => axum::http::StatusCode::UNAUTHORIZED,
+            AppError::RateLimitError(_) => axum::http::StatusCode::TOO_MANY_REQUESTS,
+            AppError::ValidationError(_) => axum::http::StatusCode::BAD_REQUEST,
+            _ => axum::http::StatusCode::INTERNAL_SERVER_ERROR,
+        };
+
+        let body = axum::Json(serde_json::json!({
+            "error": self.to_string(),
+            "type": format!("{:?}", self)
+        }));
+
+        (status, body).into_response()
+    }
+}

src/lib.rs

@@ -0,0 +1,89 @@
+//! LLM Proxy Library
+//!
+//! This library provides the core functionality for the LLM proxy gateway,
+//! including provider integration, token tracking, and API endpoints.
+
+pub mod auth;
+pub mod client;
+pub mod config;
+pub mod database;
+pub mod dashboard;
+pub mod errors;
+pub mod logging;
+pub mod models;
+pub mod multimodal;
+pub mod providers;
+pub mod rate_limiting;
+pub mod server;
+pub mod state;
+pub mod utils;
+
+// Re-exports for convenience
+pub use auth::*;
+pub use config::*;
+pub use database::*;
+pub use errors::*;
+pub use logging::*;
+pub use models::*;
+pub use providers::*;
+pub use server::*;
+pub use state::*;
+
+/// Test utilities for integration testing
+#[cfg(test)]
+pub mod test_utils {
+    use std::sync::Arc;
+    
+    use crate::{
+        state::AppState,
+        rate_limiting::RateLimitManager,
+        client::ClientManager,
+        providers::ProviderManager,
+    };
+    use sqlx::sqlite::SqlitePool;
+    
+    /// Create a test application state
+    pub async fn create_test_state() -> Arc<AppState> {
+        // Create in-memory database
+        let pool = SqlitePool::connect("sqlite::memory:")
+            .await
+            .expect("Failed to create test database");
+        
+        // Run migrations
+        crate::database::init(&crate::config::DatabaseConfig {
+            path: std::path::PathBuf::from(":memory:"),
+            max_connections: 5,
+        }).await.expect("Failed to initialize test database");
+        
+        let rate_limit_manager = RateLimitManager::new(
+            crate::rate_limiting::RateLimiterConfig::default(),
+            crate::rate_limiting::CircuitBreakerConfig::default(),
+        );
+        
+        let client_manager = Arc::new(ClientManager::new(pool.clone()));
+        
+        // Create provider manager
+        let provider_manager = ProviderManager::new();
+        
+        let model_registry = crate::models::registry::ModelRegistry {
+            providers: std::collections::HashMap::new(),
+        };
+        
+        Arc::new(AppState {
+            provider_manager,
+            db_pool: pool.clone(),
+            rate_limit_manager: Arc::new(rate_limit_manager),
+            client_manager,
+            request_logger: Arc::new(crate::logging::RequestLogger::new(pool.clone())),
+            model_registry: Arc::new(model_registry),
+        })
+    }
+    
+    /// Create a test HTTP client
+    pub fn create_test_client() -> reqwest::Client {
+        reqwest::Client::builder()
+            .timeout(std::time::Duration::from_secs(30))
+            .build()
+            .expect("Failed to create test HTTP client")
+    }
+}

src/logging/mod.rs

@@ -0,0 +1,186 @@
+use chrono::{DateTime, Utc};
+use sqlx::SqlitePool;
+use tokio::sync::broadcast;
+use tracing::warn;
+use serde::Serialize;
+
+use crate::errors::AppError;
+
+/// Request log entry for database storage
+#[derive(Debug, Clone, Serialize)]
+pub struct RequestLog {
+    pub timestamp: DateTime<Utc>,
+    pub client_id: String,
+    pub provider: String,
+    pub model: String,
+    pub prompt_tokens: u32,
+    pub completion_tokens: u32,
+    pub total_tokens: u32,
+    pub cost: f64,
+    pub has_images: bool,
+    pub status: String, // "success", "error"
+    pub error_message: Option<String>,
+    pub duration_ms: u64,
+}
+
+/// Database operations for request logging
+pub struct RequestLogger {
+    db_pool: SqlitePool,
+    dashboard_tx: broadcast::Sender<serde_json::Value>,
+}
+
+impl RequestLogger {
+    pub fn new(db_pool: SqlitePool, dashboard_tx: broadcast::Sender<serde_json::Value>) -> Self {
+        Self { db_pool, dashboard_tx }
+    }
+
+    /// Log a request to the database (async, spawns a task)
+    pub fn log_request(&self, log: RequestLog) {
+        let pool = self.db_pool.clone();
+        let tx = self.dashboard_tx.clone();
+        
+        // Spawn async task to log without blocking response
+        tokio::spawn(async move {
+            // Broadcast to dashboard
+            let _ = tx.send(serde_json::json!({
+                "event_type": "request",
+                "data": log
+            }));
+
+            if let Err(e) = Self::insert_log(&pool, log).await {
+                warn!("Failed to log request to database: {}", e);
+            }
+        });
+    }
+
+    /// Insert a log entry into the database
+    async fn insert_log(pool: &SqlitePool, log: RequestLog) -> Result<(), sqlx::Error> {
+        sqlx::query(
+            r#"
+            INSERT INTO llm_requests 
+            (timestamp, client_id, provider, model, prompt_tokens, completion_tokens, total_tokens, cost, has_images, status, error_message, duration_ms, request_body, response_body)
+            VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+            "#,
+        )
+        .bind(log.timestamp)
+        .bind(log.client_id)
+        .bind(log.provider)
+        .bind(log.model)
+        .bind(log.prompt_tokens as i64)
+        .bind(log.completion_tokens as i64)
+        .bind(log.total_tokens as i64)
+        .bind(log.cost)
+        .bind(log.has_images)
+        .bind(log.status)
+        .bind(log.error_message)
+        .bind(log.duration_ms as i64)
+        .bind(None::<String>) // request_body - TODO: store serialized request
+        .bind(None::<String>) // response_body - TODO: store serialized response or error
+        .execute(pool)
+        .await?;
+
+        Ok(())
+    }
+}
+
+// /// Middleware to log LLM API requests
+// /// TODO: Implement proper middleware that can extract response body details
+// pub async fn request_logging_middleware(
+//     // Extract the authenticated client (if available)
+//     auth_result: Result<AuthenticatedClient, AppError>,
+//     request: Request,
+//     next: Next,
+// ) -> Response {
+//     let start_time = std::time::Instant::now();
+//     
+//     // Extract client_id from auth or use "unknown"
+//     let client_id = match auth_result {
+//         Ok(auth) => auth.client_id,
+//         Err(_) => "unknown".to_string(),
+//     };
+//     
+//     // Try to extract request details
+//     let (request_parts, request_body) = request.into_parts();
+//     
+//     // Clone request parts for logging
+//     let path = request_parts.uri.path().to_string();
+//     
+//     // Check if this is a chat completion request
+//     let is_chat_completion = path == "/v1/chat/completions";
+//     
+//     // Reconstruct request for downstream handlers
+//     let request = Request::from_parts(request_parts, request_body);
+//     
+//     // Process request and get response
+//     let response = next.run(request).await;
+//     
+//     // Calculate duration
+//     let duration = start_time.elapsed();
+//     let duration_ms = duration.as_millis() as u64;
+//     
+//     // Log basic request info
+//     info!(
+//         "Request from {} to {} - Status: {} - Duration: {}ms",
+//         client_id,
+//         path,
+//         response.status().as_u16(),
+//         duration_ms
+//     );
+//     
+//     // TODO: Extract more details from request/response for logging
+//     // For now, we'll need to modify the server handler to pass additional context
+//     
+//     response
+// }
+
+/// Context for request logging that can be passed through extensions
+#[derive(Clone)]
+pub struct LoggingContext {
+    pub client_id: String,
+    pub provider_name: String,
+    pub model: String,
+    pub prompt_tokens: u32,
+    pub completion_tokens: u32,
+    pub total_tokens: u32,
+    pub cost: f64,
+    pub has_images: bool,
+    pub error: Option<AppError>,
+}
+
+impl LoggingContext {
+    pub fn new(client_id: String, provider_name: String, model: String) -> Self {
+        Self {
+            client_id,
+            provider_name,
+            model,
+            prompt_tokens: 0,
+            completion_tokens: 0,
+            total_tokens: 0,
+            cost: 0.0,
+            has_images: false,
+            error: None,
+        }
+    }
+    
+    pub fn with_token_counts(mut self, prompt_tokens: u32, completion_tokens: u32) -> Self {
+        self.prompt_tokens = prompt_tokens;
+        self.completion_tokens = completion_tokens;
+        self.total_tokens = prompt_tokens + completion_tokens;
+        self
+    }
+    
+    pub fn with_cost(mut self, cost: f64) -> Self {
+        self.cost = cost;
+        self
+    }
+    
+    pub fn with_images(mut self, has_images: bool) -> Self {
+        self.has_images = has_images;
+        self
+    }
+    
+    pub fn with_error(mut self, error: AppError) -> Self {
+        self.error = Some(error);
+        self
+    }
+}

src/main.rs

@@ -0,0 +1,141 @@
+use anyhow::Result;
+use axum::{Router, routing::get};
+use std::net::SocketAddr;
+use std::sync::Arc;
+use tracing::{info, error};
+
+use llm_proxy::{
+    config::AppConfig, 
+    state::AppState,
+    providers::{
+        ProviderManager, 
+        openai::OpenAIProvider,
+        gemini::GeminiProvider,
+        deepseek::DeepSeekProvider,
+        grok::GrokProvider,
+        ollama::OllamaProvider,
+    },
+    database,
+    server,
+    dashboard,
+    rate_limiting::{RateLimitManager, RateLimiterConfig, CircuitBreakerConfig},
+};
+
+#[tokio::main]
+async fn main() -> Result<()> {
+    // Initialize tracing (logging)
+    tracing_subscriber::fmt()
+        .with_max_level(tracing::Level::INFO)
+        .with_target(false)
+        .init();
+
+    info!("Starting LLM Proxy Gateway v{}", env!("CARGO_PKG_VERSION"));
+
+    // Load configuration
+    let config = AppConfig::load().await?;
+    info!("Configuration loaded from {:?}", config.config_path);
+
+    // Initialize database connection pool
+    let db_pool = database::init(&config.database).await?;
+    info!("Database initialized at {:?}", config.database.path);
+
+    // Initialize provider manager with configured providers
+    let mut provider_manager = ProviderManager::new();
+    
+    // Initialize OpenAI
+    if config.providers.openai.enabled {
+        match OpenAIProvider::new(&config.providers.openai, &config) {
+            Ok(p) => {
+                provider_manager.add_provider(Arc::new(p));
+                info!("OpenAI provider initialized");
+            }
+            Err(e) => error!("Failed to initialize OpenAI provider: {}", e),
+        }
+    }
+    
+    // Initialize Gemini
+    if config.providers.gemini.enabled {
+        match GeminiProvider::new(&config.providers.gemini, &config) {
+            Ok(p) => {
+                provider_manager.add_provider(Arc::new(p));
+                info!("Gemini provider initialized");
+            }
+            Err(e) => error!("Failed to initialize Gemini provider: {}", e),
+        }
+    }
+    
+    // Initialize DeepSeek
+    if config.providers.deepseek.enabled {
+        match DeepSeekProvider::new(&config.providers.deepseek, &config) {
+            Ok(p) => {
+                provider_manager.add_provider(Arc::new(p));
+                info!("DeepSeek provider initialized");
+            }
+            Err(e) => error!("Failed to initialize DeepSeek provider: {}", e),
+        }
+    }
+    
+    // Initialize Grok
+    if config.providers.grok.enabled {
+        match GrokProvider::new(&config.providers.grok, &config) {
+            Ok(p) => {
+                provider_manager.add_provider(Arc::new(p));
+                info!("Grok provider initialized");
+            }
+            Err(e) => error!("Failed to initialize Grok provider: {}", e),
+        }
+    }
+
+    // Initialize Ollama
+    if config.providers.ollama.enabled {
+        match OllamaProvider::new(&config.providers.ollama, &config) {
+            Ok(p) => {
+                provider_manager.add_provider(Arc::new(p));
+                info!("Ollama provider initialized at {}", config.providers.ollama.base_url);
+            }
+            Err(e) => error!("Failed to initialize Ollama provider: {}", e),
+        }
+    }
+
+    // Create rate limit manager
+    let rate_limit_manager = RateLimitManager::new(
+        RateLimiterConfig::default(),
+        CircuitBreakerConfig::default(),
+    );
+    
+    // Fetch model registry from models.dev
+    let model_registry = match llm_proxy::utils::registry::fetch_registry().await {
+        Ok(registry) => registry,
+        Err(e) => {
+            error!("Failed to fetch model registry: {}. Using empty registry.", e);
+            llm_proxy::models::registry::ModelRegistry { providers: std::collections::HashMap::new() }
+        }
+    };
+    
+    // Create application state
+    let state = AppState::new(provider_manager, db_pool, rate_limit_manager, model_registry);
+
+    // Create application router
+    let app = Router::new()
+        .route("/health", get(health_check))
+        .route("/", get(root))
+        .merge(server::router(state.clone()))
+        .merge(dashboard::router(state.clone()));
+
+    // Start server
+    let addr = SocketAddr::from(([0, 0, 0, 0], config.server.port));
+    info!("Server listening on http://{}", addr);
+
+    let listener = tokio::net::TcpListener::bind(&addr).await?;
+    axum::serve(listener, app).await?;
+
+    Ok(())
+}
+
+async fn health_check() -> &'static str {
+    "OK"
+}
+
+async fn root() -> &'static str {
+    "LLM Proxy Gateway - Unified interface for OpenAI, Gemini, DeepSeek, and Grok"
+}

src/models/mod.rs

@@ -0,0 +1,247 @@
+use serde::{Deserialize, Serialize};
+
+pub mod registry;
+
+// ========== OpenAI-compatible Request/Response Structs ==========
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ChatCompletionRequest {
+    pub model: String,
+    pub messages: Vec<ChatMessage>,
+    #[serde(default)]
+    pub temperature: Option<f64>,
+    #[serde(default)]
+    pub max_tokens: Option<u32>,
+    #[serde(default)]
+    pub stream: Option<bool>,
+    // Add other OpenAI-compatible fields as needed
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ChatMessage {
+    pub role: String, // "system", "user", "assistant"
+    #[serde(flatten)]
+    pub content: MessageContent,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub reasoning_content: Option<String>,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(untagged)]
+pub enum MessageContent {
+    Text { content: String },
+    Parts { content: Vec<ContentPartValue> },
+    None, // Handle cases where content might be null but reasoning is present
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(tag = "type", rename_all = "snake_case")]
+pub enum ContentPartValue {
+    Text { text: String },
+    ImageUrl { image_url: ImageUrl },
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ImageUrl {
+    pub url: String,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub detail: Option<String>,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ChatCompletionResponse {
+    pub id: String,
+    pub object: String,
+    pub created: u64,
+    pub model: String,
+    pub choices: Vec<ChatChoice>,
+    pub usage: Option<Usage>,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ChatChoice {
+    pub index: u32,
+    pub message: ChatMessage,
+    pub finish_reason: Option<String>,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct Usage {
+    pub prompt_tokens: u32,
+    pub completion_tokens: u32,
+    pub total_tokens: u32,
+}
+
+// ========== Streaming Response Structs ==========
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ChatCompletionStreamResponse {
+    pub id: String,
+    pub object: String,
+    pub created: u64,
+    pub model: String,
+    pub choices: Vec<ChatStreamChoice>,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ChatStreamChoice {
+    pub index: u32,
+    pub delta: ChatStreamDelta,
+    pub finish_reason: Option<String>,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ChatStreamDelta {
+    pub role: Option<String>,
+    pub content: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub reasoning_content: Option<String>,
+}
+
+// ========== Unified Request Format (for internal use) ==========
+
+#[derive(Debug, Clone)]
+pub struct UnifiedRequest {
+    pub client_id: String,
+    pub model: String,
+    pub messages: Vec<UnifiedMessage>,
+    pub temperature: Option<f64>,
+    pub max_tokens: Option<u32>,
+    pub stream: bool,
+    pub has_images: bool,
+}
+
+#[derive(Debug, Clone)]
+pub struct UnifiedMessage {
+    pub role: String,
+    pub content: Vec<ContentPart>,
+}
+
+#[derive(Debug, Clone)]
+pub enum ContentPart {
+    Text { text: String },
+    Image(crate::multimodal::ImageInput),
+}
+
+// ========== Provider-specific Structs ==========
+
+#[derive(Debug, Clone, Serialize)]
+pub struct OpenAIRequest {
+    pub model: String,
+    pub messages: Vec<OpenAIMessage>,
+    pub temperature: Option<f64>,
+    pub max_tokens: Option<u32>,
+    pub stream: Option<bool>,
+}
+
+#[derive(Debug, Clone, Serialize)]
+pub struct OpenAIMessage {
+    pub role: String,
+    pub content: Vec<OpenAIContentPart>,
+}
+
+#[derive(Debug, Clone, Serialize)]
+#[serde(tag = "type", rename_all = "snake_case")]
+pub enum OpenAIContentPart {
+    Text { text: String },
+    ImageUrl { image_url: ImageUrl },
+}
+
+// Note: ImageUrl struct is defined earlier in the file
+
+// ========== Conversion Traits ==========
+
+pub trait ToOpenAI {
+    fn to_openai(&self) -> Result<OpenAIRequest, anyhow::Error>;
+}
+
+pub trait FromOpenAI {
+    fn from_openai(request: &OpenAIRequest) -> Result<Self, anyhow::Error>
+    where
+        Self: Sized;
+}
+
+impl UnifiedRequest {
+    /// Hydrate all image content by fetching URLs and converting to base64/bytes
+    pub async fn hydrate_images(&mut self) -> anyhow::Result<()> {
+        if !self.has_images {
+            return Ok(());
+        }
+
+        for msg in &mut self.messages {
+            for part in &mut msg.content {
+                if let ContentPart::Image(image_input) = part {
+                    // Pre-fetch and validate if it's a URL
+                    if let crate::multimodal::ImageInput::Url(_url) = image_input {
+                        let (base64_data, mime_type) = image_input.to_base64().await?;
+                        *image_input = crate::multimodal::ImageInput::Base64 {
+                            data: base64_data,
+                            mime_type,
+                        };
+                    }
+                }
+            }
+        }
+        Ok(())
+    }
+}
+
+impl TryFrom<ChatCompletionRequest> for UnifiedRequest {
+    type Error = anyhow::Error;
+
+    fn try_from(req: ChatCompletionRequest) -> Result<Self, Self::Error> {
+        let mut has_images = false;
+
+        // Convert OpenAI-compatible request to unified format
+        let messages = req
+            .messages
+            .into_iter()
+            .map(|msg| {
+                let (content, _images_in_message) = match msg.content {
+                    MessageContent::Text { content } => {
+                        (vec![ContentPart::Text { text: content }], false)
+                    }
+                    MessageContent::Parts { content } => {
+                        let mut unified_content = Vec::new();
+                        let mut has_images_in_msg = false;
+
+                        for part in content {
+                            match part {
+                                ContentPartValue::Text { text } => {
+                                    unified_content.push(ContentPart::Text { text });
+                                }
+                                ContentPartValue::ImageUrl { image_url } => {
+                                    has_images_in_msg = true;
+                                    has_images = true;
+                                    unified_content.push(ContentPart::Image(
+                                        crate::multimodal::ImageInput::from_url(image_url.url)
+                                    ));
+                                }
+                            }
+                        }
+
+                        (unified_content, has_images_in_msg)
+                    }
+                    MessageContent::None => {
+                        (vec![], false)
+                    }
+                };
+
+                UnifiedMessage {
+                    role: msg.role,
+                    content,
+                }
+            })
+            .collect();
+
+        Ok(UnifiedRequest {
+            client_id: String::new(), // Will be populated by auth middleware
+            model: req.model,
+            messages,
+            temperature: req.temperature,
+            max_tokens: req.max_tokens,
+            stream: req.stream.unwrap_or(false),
+            has_images,
+        })
+    }
+}

src/models/registry.rs

@@ -0,0 +1,69 @@
+use serde::{Deserialize, Serialize};
+use std::collections::HashMap;
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ModelRegistry {
+    #[serde(flatten)]
+    pub providers: HashMap<String, ProviderInfo>,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ProviderInfo {
+    pub id: String,
+    pub name: String,
+    pub models: HashMap<String, ModelMetadata>,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ModelMetadata {
+    pub id: String,
+    pub name: String,
+    pub cost: Option<ModelCost>,
+    pub limit: Option<ModelLimit>,
+    pub modalities: Option<ModelModalities>,
+    pub tool_call: Option<bool>,
+    pub reasoning: Option<bool>,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ModelCost {
+    pub input: f64,
+    pub output: f64,
+    pub cache_read: Option<f64>,
+    pub cache_write: Option<f64>,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ModelLimit {
+    pub context: u32,
+    pub output: u32,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ModelModalities {
+    pub input: Vec<String>,
+    pub output: Vec<String>,
+}
+
+impl ModelRegistry {
+    /// Find a model by its ID (searching across all providers)
+    pub fn find_model(&self, model_id: &str) -> Option<&ModelMetadata> {
+        // First try exact match if the key in models map matches the ID
+        for provider in self.providers.values() {
+            if let Some(model) = provider.models.get(model_id) {
+                return Some(model);
+            }
+        }
+        
+        // Try searching for the model ID inside the metadata if the key was different
+        for provider in self.providers.values() {
+            for model in provider.models.values() {
+                if model.id == model_id {
+                    return Some(model);
+                }
+            }
+        }
+        
+        None
+    }
+}

src/multimodal/mod.rs

@@ -0,0 +1,285 @@
+//! Multimodal support for image processing and conversion
+//! 
+//! This module handles:
+//! 1. Image format detection and conversion
+//! 2. Base64 encoding/decoding
+//! 3. URL fetching for images
+//! 4. Provider-specific image format conversion
+
+use anyhow::{Context, Result};
+use base64::{engine::general_purpose, Engine as _};
+use tracing::{info, warn};
+
+/// Supported image formats for multimodal input
+#[derive(Debug, Clone)]
+pub enum ImageInput {
+    /// Base64-encoded image data with MIME type
+    Base64 {
+        data: String,
+        mime_type: String,
+    },
+    /// URL to fetch image from
+    Url(String),
+    /// Raw bytes with MIME type
+    Bytes {
+        data: Vec<u8>,
+        mime_type: String,
+    },
+}
+
+impl ImageInput {
+    /// Create ImageInput from base64 string
+    pub fn from_base64(data: String, mime_type: String) -> Self {
+        Self::Base64 { data, mime_type }
+    }
+    
+    /// Create ImageInput from URL
+    pub fn from_url(url: String) -> Self {
+        Self::Url(url)
+    }
+    
+    /// Create ImageInput from raw bytes
+    pub fn from_bytes(data: Vec<u8>, mime_type: String) -> Self {
+        Self::Bytes { data, mime_type }
+    }
+    
+    /// Get MIME type if available
+    pub fn mime_type(&self) -> Option<&str> {
+        match self {
+            Self::Base64 { mime_type, .. } => Some(mime_type),
+            Self::Bytes { mime_type, .. } => Some(mime_type),
+            Self::Url(_) => None,
+        }
+    }
+    
+    /// Convert to base64 if not already
+    pub async fn to_base64(&self) -> Result<(String, String)> {
+        match self {
+            Self::Base64 { data, mime_type } => Ok((data.clone(), mime_type.clone())),
+            Self::Bytes { data, mime_type } => {
+                let base64_data = general_purpose::STANDARD.encode(data);
+                Ok((base64_data, mime_type.clone()))
+            }
+            Self::Url(url) => {
+                // Fetch image from URL
+                info!("Fetching image from URL: {}", url);
+                let response = reqwest::get(url)
+                    .await
+                    .context("Failed to fetch image from URL")?;
+                
+                if !response.status().is_success() {
+                    anyhow::bail!("Failed to fetch image: HTTP {}", response.status());
+                }
+                
+                let mime_type = response
+                    .headers()
+                    .get(reqwest::header::CONTENT_TYPE)
+                    .and_then(|h| h.to_str().ok())
+                    .unwrap_or("image/jpeg")
+                    .to_string();
+                
+                let bytes = response.bytes().await.context("Failed to read image bytes")?;
+                
+                let base64_data = general_purpose::STANDARD.encode(&bytes);
+                Ok((base64_data, mime_type))
+            }
+        }
+    }
+    
+    /// Get image dimensions (width, height)
+    pub async fn get_dimensions(&self) -> Result<(u32, u32)> {
+        let bytes = match self {
+            Self::Base64 { data, .. } => {
+                general_purpose::STANDARD.decode(data).context("Failed to decode base64")?
+            }
+            Self::Bytes { data, .. } => data.clone(),
+            Self::Url(_) => {
+                let (base64_data, _) = self.to_base64().await?;
+                general_purpose::STANDARD.decode(&base64_data).context("Failed to decode base64")?
+            }
+        };
+        
+        let img = image::load_from_memory(&bytes).context("Failed to load image from bytes")?;
+        Ok((img.width(), img.height()))
+    }
+    
+    /// Validate image size and format
+    pub async fn validate(&self, max_size_mb: f64) -> Result<()> {
+        let (width, height) = self.get_dimensions().await?;
+        
+        // Check dimensions
+        if width > 4096 || height > 4096 {
+            warn!("Image dimensions too large: {}x{}", width, height);
+            // Continue anyway, but log warning
+        }
+        
+        // Check file size
+        let size_bytes = match self {
+            Self::Base64 { data, .. } => {
+                // Base64 size is ~4/3 of original
+                (data.len() as f64 * 0.75) as usize
+            }
+            Self::Bytes { data, .. } => data.len(),
+            Self::Url(_) => {
+                // For URLs, we'd need to fetch to check size
+                // Skip size check for URLs for now
+                return Ok(());
+            }
+        };
+        
+        let size_mb = size_bytes as f64 / (1024.0 * 1024.0);
+        if size_mb > max_size_mb {
+            anyhow::bail!("Image too large: {:.2}MB > {:.2}MB limit", size_mb, max_size_mb);
+        }
+        
+        Ok(())
+    }
+}
+
+/// Provider-specific image format conversion
+pub struct ImageConverter;
+
+impl ImageConverter {
+    /// Convert image to OpenAI-compatible format
+    pub async fn to_openai_format(image: &ImageInput) -> Result<serde_json::Value> {
+        let (base64_data, mime_type) = image.to_base64().await?;
+        
+        // OpenAI expects data URL format: "data:image/jpeg;base64,{data}"
+        let data_url = format!("data:{};base64,{}", mime_type, base64_data);
+        
+        Ok(serde_json::json!({
+            "type": "image_url",
+            "image_url": {
+                "url": data_url,
+                "detail": "auto"  // Can be "low", "high", or "auto"
+            }
+        }))
+    }
+    
+    /// Convert image to Gemini-compatible format
+    pub async fn to_gemini_format(image: &ImageInput) -> Result<serde_json::Value> {
+        let (base64_data, mime_type) = image.to_base64().await?;
+        
+        // Gemini expects inline data format
+        Ok(serde_json::json!({
+            "inline_data": {
+                "mime_type": mime_type,
+                "data": base64_data
+            }
+        }))
+    }
+    
+    /// Convert image to DeepSeek-compatible format
+    pub async fn to_deepseek_format(image: &ImageInput) -> Result<serde_json::Value> {
+        // DeepSeek uses OpenAI-compatible format for vision models
+        Self::to_openai_format(image).await
+    }
+    
+    /// Detect if a model supports multimodal input
+    pub fn model_supports_multimodal(model: &str) -> bool {
+        // OpenAI vision models
+        if (model.starts_with("gpt-4") && (model.contains("vision") || model.contains("-v") || model.contains("4o"))) || 
+           model.starts_with("o1-") || model.starts_with("o3-") {
+            return true;
+        }
+        
+        // Gemini vision models
+        if model.starts_with("gemini") {
+            // Most Gemini models support vision
+            return true;
+        }
+        
+        // DeepSeek vision models
+        if model.starts_with("deepseek-vl") {
+            return true;
+        }
+        
+        false
+    }
+}
+
+/// Parse OpenAI-compatible multimodal message content
+pub fn parse_openai_content(content: &serde_json::Value) -> Result<Vec<(String, Option<ImageInput>)>> {
+    let mut parts = Vec::new();
+    
+    if let Some(content_str) = content.as_str() {
+        // Simple text content
+        parts.push((content_str.to_string(), None));
+    } else if let Some(content_array) = content.as_array() {
+        // Array of content parts (text and/or images)
+        for part in content_array {
+            if let Some(part_obj) = part.as_object() {
+                if let Some(part_type) = part_obj.get("type").and_then(|t| t.as_str()) {
+                    match part_type {
+                        "text" => {
+                            if let Some(text) = part_obj.get("text").and_then(|t| t.as_str()) {
+                                parts.push((text.to_string(), None));
+                            }
+                        }
+                        "image_url" => {
+                            if let Some(image_url_obj) = part_obj.get("image_url").and_then(|o| o.as_object()) {
+                                if let Some(url) = image_url_obj.get("url").and_then(|u| u.as_str()) {
+                                    if url.starts_with("data:") {
+                                        // Parse data URL
+                                        if let Some((mime_type, data)) = parse_data_url(url) {
+                                            let image_input = ImageInput::from_base64(data, mime_type);
+                                            parts.push(("".to_string(), Some(image_input)));
+                                        }
+                                    } else {
+                                        // Regular URL
+                                        let image_input = ImageInput::from_url(url.to_string());
+                                        parts.push(("".to_string(), Some(image_input)));
+                                    }
+                                }
+                            }
+                        }
+                        _ => {
+                            warn!("Unknown content part type: {}", part_type);
+                        }
+                    }
+                }
+            }
+        }
+    }
+    
+    Ok(parts)
+}
+
+/// Parse data URL (data:image/jpeg;base64,{data})
+fn parse_data_url(data_url: &str) -> Option<(String, String)> {
+    if !data_url.starts_with("data:") {
+        return None;
+    }
+    
+    let parts: Vec<&str> = data_url[5..].split(";base64,").collect();
+    if parts.len() != 2 {
+        return None;
+    }
+    
+    let mime_type = parts[0].to_string();
+    let data = parts[1].to_string();
+    
+    Some((mime_type, data))
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    
+    #[tokio::test]
+    async fn test_parse_data_url() {
+        let test_url = "data:image/jpeg;base64,SGVsbG8gV29ybGQ="; // "Hello World" in base64
+        let (mime_type, data) = parse_data_url(test_url).unwrap();
+        
+        assert_eq!(mime_type, "image/jpeg");
+        assert_eq!(data, "SGVsbG8gV29ybGQ=");
+    }
+    
+    #[tokio::test]
+    async fn test_model_supports_multimodal() {
+        assert!(ImageConverter::model_supports_multimodal("gpt-4-vision-preview"));
+        assert!(ImageConverter::model_supports_multimodal("gemini-pro-vision"));
+        assert!(!ImageConverter::model_supports_multimodal("gpt-3.5-turbo"));
+        assert!(!ImageConverter::model_supports_multimodal("gemini-pro"));
+    }
+}

src/providers/deepseek.rs

@@ -0,0 +1,209 @@
+use async_trait::async_trait;
+use anyhow::Result;
+use futures::stream::{BoxStream, StreamExt};
+use serde_json::Value;
+
+use crate::{
+    models::UnifiedRequest,
+    errors::AppError,
+    config::AppConfig,
+};
+use super::{ProviderResponse, ProviderStreamChunk};
+
+pub struct DeepSeekProvider {
+    client: reqwest::Client,
+    config: crate::config::DeepSeekConfig,
+    api_key: String,
+    pricing: Vec<crate::config::ModelPricing>,
+}
+
+impl DeepSeekProvider {
+    pub fn new(config: &crate::config::DeepSeekConfig, app_config: &AppConfig) -> Result<Self> {
+        let api_key = app_config.get_api_key("deepseek")?;
+        
+        Ok(Self {
+            client: reqwest::Client::new(),
+            config: config.clone(),
+            api_key,
+            pricing: app_config.pricing.deepseek.clone(),
+        })
+    }
+}
+
+#[async_trait]
+impl super::Provider for DeepSeekProvider {
+    fn name(&self) -> &str {
+        "deepseek"
+    }
+
+    fn supports_model(&self, model: &str) -> bool {
+        model.starts_with("deepseek-") || model.contains("deepseek")
+    }
+
+    fn supports_multimodal(&self) -> bool {
+        false
+    }
+
+    async fn chat_completion(
+        &self,
+        request: UnifiedRequest,
+    ) -> Result<ProviderResponse, AppError> {
+        // Build the OpenAI-compatible body
+        let mut body = serde_json::json!({
+            "model": request.model,
+            "messages": request.messages.iter().map(|m| {
+                serde_json::json!({
+                    "role": m.role,
+                    "content": m.content.iter().map(|p| {
+                        match p {
+                            crate::models::ContentPart::Text { text } => serde_json::json!({ "type": "text", "text": text }),
+                            crate::models::ContentPart::Image(image_input) => {
+                                // DeepSeek currently doesn't support images in the same way, but we'll try to be standard
+                                let (base64_data, mime_type) = futures::executor::block_on(image_input.to_base64()).unwrap_or_default();
+                                serde_json::json!({
+                                    "type": "image_url",
+                                    "image_url": { "url": format!("data:{};base64,{}", mime_type, base64_data) }
+                                })
+                            }
+                        }
+                    }).collect::<Vec<_>>()
+                })
+            }).collect::<Vec<_>>(),
+            "stream": false,
+        });
+
+        if let Some(temp) = request.temperature {
+            body["temperature"] = serde_json::json!(temp);
+        }
+        if let Some(max_tokens) = request.max_tokens {
+            body["max_tokens"] = serde_json::json!(max_tokens);
+        }
+
+        let response = self.client.post(format!("{}/chat/completions", self.config.base_url))
+            .header("Authorization", format!("Bearer {}", self.api_key))
+            .json(&body)
+            .send()
+            .await
+            .map_err(|e| AppError::ProviderError(e.to_string()))?;
+
+        if !response.status().is_success() {
+            let error_text = response.text().await.unwrap_or_default();
+            return Err(AppError::ProviderError(format!("DeepSeek API error: {}", error_text)));
+        }
+
+        let resp_json: Value = response.json().await.map_err(|e| AppError::ProviderError(e.to_string()))?;
+        
+        let choice = resp_json["choices"].get(0).ok_or_else(|| AppError::ProviderError("No choices in response".to_string()))?;
+        let message = &choice["message"];
+        
+        let content = message["content"].as_str().unwrap_or_default().to_string();
+        let reasoning_content = message["reasoning_content"].as_str().map(|s| s.to_string());
+        
+        let usage = &resp_json["usage"];
+        let prompt_tokens = usage["prompt_tokens"].as_u64().unwrap_or(0) as u32;
+        let completion_tokens = usage["completion_tokens"].as_u64().unwrap_or(0) as u32;
+        let total_tokens = usage["total_tokens"].as_u64().unwrap_or(0) as u32;
+
+        Ok(ProviderResponse {
+            content,
+            reasoning_content,
+            prompt_tokens,
+            completion_tokens,
+            total_tokens,
+            model: request.model,
+        })
+    }
+
+    fn estimate_tokens(&self, request: &UnifiedRequest) -> Result<u32> {
+        Ok(crate::utils::tokens::estimate_request_tokens(&request.model, request))
+    }
+
+    fn calculate_cost(&self, model: &str, prompt_tokens: u32, completion_tokens: u32, registry: &crate::models::registry::ModelRegistry) -> f64 {
+        if let Some(metadata) = registry.find_model(model) {
+            if let Some(cost) = &metadata.cost {
+                return (prompt_tokens as f64 * cost.input / 1_000_000.0) + 
+                       (completion_tokens as f64 * cost.output / 1_000_000.0);
+            }
+        }
+        
+        let (prompt_rate, completion_rate) = self.pricing.iter()
+            .find(|p| model.contains(&p.model))
+            .map(|p| (p.prompt_tokens_per_million, p.completion_tokens_per_million))
+            .unwrap_or((0.14, 0.28));
+            
+        (prompt_tokens as f64 * prompt_rate / 1_000_000.0) + (completion_tokens as f64 * completion_rate / 1_000_000.0)
+    }
+
+    async fn chat_completion_stream(
+        &self,
+        request: UnifiedRequest,
+    ) -> Result<BoxStream<'static, Result<ProviderStreamChunk, AppError>>, AppError> {
+        let mut body = serde_json::json!({
+            "model": request.model,
+            "messages": request.messages.iter().map(|m| {
+                serde_json::json!({
+                    "role": m.role,
+                    "content": m.content.iter().map(|p| {
+                        match p {
+                            crate::models::ContentPart::Text { text } => serde_json::json!({ "type": "text", "text": text }),
+                            crate::models::ContentPart::Image(_) => serde_json::json!({ "type": "text", "text": "[Image]" }),
+                        }
+                    }).collect::<Vec<_>>()
+                })
+            }).collect::<Vec<_>>(),
+            "stream": true,
+        });
+
+        if let Some(temp) = request.temperature {
+            body["temperature"] = serde_json::json!(temp);
+        }
+        if let Some(max_tokens) = request.max_tokens {
+            body["max_tokens"] = serde_json::json!(max_tokens);
+        }
+
+        // Create eventsource stream
+        use reqwest_eventsource::{EventSource, Event};
+        let es = EventSource::new(self.client.post(format!("{}/chat/completions", self.config.base_url))
+            .header("Authorization", format!("Bearer {}", self.api_key))
+            .json(&body))
+            .map_err(|e| AppError::ProviderError(format!("Failed to create EventSource: {}", e)))?;
+        
+        let model = request.model.clone();
+        
+        let stream = async_stream::try_stream! {
+            let mut es = es;
+            while let Some(event) = es.next().await {
+                match event {
+                    Ok(Event::Message(msg)) => {
+                        if msg.data == "[DONE]" {
+                            break;
+                        }
+                        
+                        let chunk: Value = serde_json::from_str(&msg.data)
+                            .map_err(|e| AppError::ProviderError(format!("Failed to parse stream chunk: {}", e)))?;
+                        
+                        if let Some(choice) = chunk["choices"].get(0) {
+                            let delta = &choice["delta"];
+                            let content = delta["content"].as_str().unwrap_or_default().to_string();
+                            let reasoning_content = delta["reasoning_content"].as_str().map(|s| s.to_string());
+                            let finish_reason = choice["finish_reason"].as_str().map(|s| s.to_string());
+                            
+                            yield ProviderStreamChunk {
+                                content,
+                                reasoning_content,
+                                finish_reason,
+                                model: model.clone(),
+                            };
+                        }
+                    }
+                    Ok(_) => continue,
+                    Err(e) => {
+                        Err(AppError::ProviderError(format!("Stream error: {}", e)))?;
+                    }
+                }
+            }
+        };
+        
+        Ok(Box::pin(stream))
+    }
+}

src/providers/gemini.rs

@@ -0,0 +1,344 @@
+use async_trait::async_trait;
+use anyhow::Result;
+use serde::{Deserialize, Serialize};
+use futures::stream::BoxStream;
+
+use crate::{
+    models::UnifiedRequest,
+    errors::AppError,
+    config::AppConfig,
+};
+use super::{ProviderResponse, ProviderStreamChunk};
+
+#[derive(Debug, Serialize)]
+struct GeminiRequest {
+    contents: Vec<GeminiContent>,
+    generation_config: Option<GeminiGenerationConfig>,
+}
+
+#[derive(Debug, Serialize, Deserialize)]
+struct GeminiContent {
+    parts: Vec<GeminiPart>,
+    role: String,
+}
+
+#[derive(Debug, Serialize, Deserialize)]
+struct GeminiPart {
+    #[serde(skip_serializing_if = "Option::is_none")]
+    text: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    inline_data: Option<GeminiInlineData>,
+}
+
+#[derive(Debug, Serialize, Deserialize)]
+struct GeminiInlineData {
+    mime_type: String,
+    data: String,
+}
+
+#[derive(Debug, Serialize)]
+struct GeminiGenerationConfig {
+    temperature: Option<f64>,
+    max_output_tokens: Option<u32>,
+}
+
+#[derive(Debug, Deserialize)]
+struct GeminiCandidate {
+    content: GeminiContent,
+    _finish_reason: Option<String>,
+}
+
+#[derive(Debug, Deserialize)]
+struct GeminiUsageMetadata {
+    prompt_token_count: u32,
+    candidates_token_count: u32,
+    total_token_count: u32,
+}
+
+#[derive(Debug, Deserialize)]
+struct GeminiResponse {
+    candidates: Vec<GeminiCandidate>,
+    usage_metadata: Option<GeminiUsageMetadata>,
+}
+
+
+
+pub struct GeminiProvider {
+    client: reqwest::Client,
+    config: crate::config::GeminiConfig,
+    api_key: String,
+    pricing: Vec<crate::config::ModelPricing>,
+}
+
+impl GeminiProvider {
+    pub fn new(config: &crate::config::GeminiConfig, app_config: &AppConfig) -> Result<Self> {
+        let api_key = app_config.get_api_key("gemini")?;
+        
+        let client = reqwest::Client::builder()
+            .timeout(std::time::Duration::from_secs(30))
+            .build()?;
+        
+        Ok(Self {
+            client,
+            config: config.clone(),
+            api_key,
+            pricing: app_config.pricing.gemini.clone(),
+        })
+    }
+}
+
+#[async_trait]
+impl super::Provider for GeminiProvider {
+    fn name(&self) -> &str {
+        "gemini"
+    }
+
+    fn supports_model(&self, model: &str) -> bool {
+        model.starts_with("gemini-")
+    }
+
+    fn supports_multimodal(&self) -> bool {
+        true  // Gemini supports vision
+    }
+
+    async fn chat_completion(
+        &self,
+        request: UnifiedRequest,
+    ) -> Result<ProviderResponse, AppError> {
+        // Convert UnifiedRequest to Gemini request
+        let mut contents = Vec::with_capacity(request.messages.len());
+        
+        for msg in request.messages {
+            let mut parts = Vec::with_capacity(msg.content.len());
+            
+            for part in msg.content {
+                match part {
+                    crate::models::ContentPart::Text { text } => {
+                        parts.push(GeminiPart {
+                            text: Some(text),
+                            inline_data: None,
+                        });
+                    }
+                    crate::models::ContentPart::Image(image_input) => {
+                        let (base64_data, mime_type) = image_input.to_base64().await
+                            .map_err(|e| AppError::ProviderError(format!("Failed to convert image: {}", e)))?;
+                        
+                        parts.push(GeminiPart {
+                            text: None,
+                            inline_data: Some(GeminiInlineData {
+                                mime_type,
+                                data: base64_data,
+                            }),
+                        });
+                    }
+                }
+            }
+            
+            // Map role: "user" -> "user", "assistant" -> "model", "system" -> "user"
+            let role = match msg.role.as_str() {
+                "assistant" => "model".to_string(),
+                _ => "user".to_string(),
+            };
+            
+            contents.push(GeminiContent {
+                parts,
+                role,
+            });
+        }
+        
+        if contents.is_empty() {
+            return Err(AppError::ProviderError("No valid text messages to send".to_string()));
+        }
+        
+        // Build generation config
+        let generation_config = if request.temperature.is_some() || request.max_tokens.is_some() {
+            Some(GeminiGenerationConfig {
+                temperature: request.temperature,
+                max_output_tokens: request.max_tokens,
+            })
+        } else {
+            None
+        };
+        
+        let gemini_request = GeminiRequest {
+            contents,
+            generation_config,
+        };
+        
+        // Build URL
+        let url = format!("{}/models/{}:generateContent?key={}", 
+            self.config.base_url, 
+            request.model,
+            self.api_key
+        );
+        
+        // Send request
+        let response = self.client
+            .post(&url)
+            .json(&gemini_request)
+            .send()
+            .await
+            .map_err(|e| AppError::ProviderError(format!("HTTP request failed: {}", e)))?;
+        
+        // Check status
+        let status = response.status();
+        if !status.is_success() {
+            let error_text = response.text().await.unwrap_or_default();
+            return Err(AppError::ProviderError(format!("Gemini API error ({}): {}", status, error_text)));
+        }
+        
+        let gemini_response: GeminiResponse = response
+            .json()
+            .await
+            .map_err(|e| AppError::ProviderError(format!("Failed to parse response: {}", e)))?;
+        
+        // Extract content from first candidate
+        let content = gemini_response.candidates
+            .first()
+            .and_then(|c| c.content.parts.first())
+            .and_then(|p| p.text.clone())
+            .unwrap_or_default();
+        
+        // Extract token usage
+        let prompt_tokens = gemini_response.usage_metadata.as_ref().map(|u| u.prompt_token_count).unwrap_or(0);
+        let completion_tokens = gemini_response.usage_metadata.as_ref().map(|u| u.candidates_token_count).unwrap_or(0);
+        let total_tokens = gemini_response.usage_metadata.as_ref().map(|u| u.total_token_count).unwrap_or(0);
+        
+        Ok(ProviderResponse {
+            content,
+            reasoning_content: None, // Gemini doesn't use this field name
+            prompt_tokens,
+            completion_tokens,
+            total_tokens,
+            model: request.model,
+        })
+    }
+
+    fn estimate_tokens(&self, request: &UnifiedRequest) -> Result<u32> {
+        Ok(crate::utils::tokens::estimate_request_tokens(&request.model, request))
+    }
+
+    fn calculate_cost(&self, model: &str, prompt_tokens: u32, completion_tokens: u32, registry: &crate::models::registry::ModelRegistry) -> f64 {
+        if let Some(metadata) = registry.find_model(model) {
+            if let Some(cost) = &metadata.cost {
+                return (prompt_tokens as f64 * cost.input / 1_000_000.0) + 
+                       (completion_tokens as f64 * cost.output / 1_000_000.0);
+            }
+        }
+        
+        let (prompt_rate, completion_rate) = self.pricing.iter()
+            .find(|p| model.contains(&p.model))
+            .map(|p| (p.prompt_tokens_per_million, p.completion_tokens_per_million))
+            .unwrap_or((0.075, 0.30)); // Default to Gemini 2.0 Flash price if not found
+            
+        (prompt_tokens as f64 * prompt_rate / 1_000_000.0) + (completion_tokens as f64 * completion_rate / 1_000_000.0)
+    }
+
+    async fn chat_completion_stream(
+        &self,
+        request: UnifiedRequest,
+    ) -> Result<BoxStream<'static, Result<ProviderStreamChunk, AppError>>, AppError> {
+        // Convert UnifiedRequest to Gemini request
+        let mut contents = Vec::with_capacity(request.messages.len());
+        
+        for msg in request.messages {
+            let mut parts = Vec::with_capacity(msg.content.len());
+            
+            for part in msg.content {
+                match part {
+                    crate::models::ContentPart::Text { text } => {
+                        parts.push(GeminiPart {
+                            text: Some(text),
+                            inline_data: None,
+                        });
+                    }
+                    crate::models::ContentPart::Image(image_input) => {
+                        let (base64_data, mime_type) = image_input.to_base64().await
+                            .map_err(|e| AppError::ProviderError(format!("Failed to convert image: {}", e)))?;
+                        
+                        parts.push(GeminiPart {
+                            text: None,
+                            inline_data: Some(GeminiInlineData {
+                                mime_type,
+                                data: base64_data,
+                            }),
+                        });
+                    }
+                }
+            }
+            
+            // Map role
+            let role = match msg.role.as_str() {
+                "assistant" => "model".to_string(),
+                _ => "user".to_string(),
+            };
+            
+            contents.push(GeminiContent {
+                parts,
+                role,
+            });
+        }
+        
+        // Build generation config
+        let generation_config = if request.temperature.is_some() || request.max_tokens.is_some() {
+            Some(GeminiGenerationConfig {
+                temperature: request.temperature,
+                max_output_tokens: request.max_tokens,
+            })
+        } else {
+            None
+        };
+        
+        let gemini_request = GeminiRequest {
+            contents,
+            generation_config,
+        };
+        
+        // Build URL for streaming
+        let url = format!("{}/models/{}:streamGenerateContent?alt=sse&key={}", 
+            self.config.base_url,
+            request.model,
+            self.api_key
+        );
+        
+        // Create eventsource stream
+        use reqwest_eventsource::{EventSource, Event};
+        use futures::StreamExt;
+
+        let es = EventSource::new(self.client.post(&url).json(&gemini_request))
+            .map_err(|e| AppError::ProviderError(format!("Failed to create EventSource: {}", e)))?;
+        
+        let model = request.model.clone();
+        
+        let stream = async_stream::try_stream! {
+            let mut es = es;
+            while let Some(event) = es.next().await {
+                match event {
+                    Ok(Event::Message(msg)) => {
+                        let gemini_response: GeminiResponse = serde_json::from_str(&msg.data)
+                            .map_err(|e| AppError::ProviderError(format!("Failed to parse stream chunk: {}", e)))?;
+                        
+                        if let Some(candidate) = gemini_response.candidates.first() {
+                            let content = candidate.content.parts.first()
+                                .and_then(|p| p.text.clone())
+                                .unwrap_or_default();
+                            
+                            yield ProviderStreamChunk {
+                                content,
+                                reasoning_content: None,
+                                finish_reason: None, // Will be set in the last chunk
+                                model: model.clone(),
+                            };
+                        }
+                    }
+                    Ok(_) => continue,
+                    Err(e) => {
+                        Err(AppError::ProviderError(format!("Stream error: {}", e)))?;
+                    }
+                }
+            }
+        };
+        
+        Ok(Box::pin(stream))
+    }
+}

src/providers/grok.rs

@@ -0,0 +1,213 @@
+use async_trait::async_trait;
+use anyhow::Result;
+use futures::stream::{BoxStream, StreamExt};
+use serde_json::Value;
+
+use crate::{
+    models::UnifiedRequest,
+    errors::AppError,
+    config::AppConfig,
+};
+use super::{ProviderResponse, ProviderStreamChunk};
+
+pub struct GrokProvider {
+    client: reqwest::Client,
+    _config: crate::config::GrokConfig,
+    api_key: String,
+    pricing: Vec<crate::config::ModelPricing>,
+}
+
+impl GrokProvider {
+    pub fn new(config: &crate::config::GrokConfig, app_config: &AppConfig) -> Result<Self> {
+        let api_key = app_config.get_api_key("grok")?;
+        
+        Ok(Self {
+            client: reqwest::Client::new(),
+            _config: config.clone(),
+            api_key,
+            pricing: app_config.pricing.grok.clone(),
+        })
+    }
+}
+
+#[async_trait]
+impl super::Provider for GrokProvider {
+    fn name(&self) -> &str {
+        "grok"
+    }
+
+    fn supports_model(&self, model: &str) -> bool {
+        model.starts_with("grok-")
+    }
+
+    fn supports_multimodal(&self) -> bool {
+        true
+    }
+
+    async fn chat_completion(
+        &self,
+        request: UnifiedRequest,
+    ) -> Result<ProviderResponse, AppError> {
+        let mut body = serde_json::json!({
+            "model": request.model,
+            "messages": request.messages.iter().map(|m| {
+                serde_json::json!({
+                    "role": m.role,
+                    "content": m.content.iter().map(|p| {
+                        match p {
+                            crate::models::ContentPart::Text { text } => serde_json::json!({ "type": "text", "text": text }),
+                            crate::models::ContentPart::Image(image_input) => {
+                                let (base64_data, mime_type) = futures::executor::block_on(image_input.to_base64()).unwrap_or_default();
+                                serde_json::json!({
+                                    "type": "image_url",
+                                    "image_url": { "url": format!("data:{};base64,{}", mime_type, base64_data) }
+                                })
+                            }
+                        }
+                    }).collect::<Vec<_>>()
+                })
+            }).collect::<Vec<_>>(),
+            "stream": false,
+        });
+
+        if let Some(temp) = request.temperature {
+            body["temperature"] = serde_json::json!(temp);
+        }
+        if let Some(max_tokens) = request.max_tokens {
+            body["max_tokens"] = serde_json::json!(max_tokens);
+        }
+
+        let response = self.client.post(format!("{}/chat/completions", self._config.base_url))
+            .header("Authorization", format!("Bearer {}", self.api_key))
+            .json(&body)
+            .send()
+            .await
+            .map_err(|e| AppError::ProviderError(e.to_string()))?;
+
+        if !response.status().is_success() {
+            let error_text = response.text().await.unwrap_or_default();
+            return Err(AppError::ProviderError(format!("Grok API error: {}", error_text)));
+        }
+
+        let resp_json: Value = response.json().await.map_err(|e| AppError::ProviderError(e.to_string()))?;
+        
+        let choice = resp_json["choices"].get(0).ok_or_else(|| AppError::ProviderError("No choices in response".to_string()))?;
+        let message = &choice["message"];
+        
+        let content = message["content"].as_str().unwrap_or_default().to_string();
+        let reasoning_content = message["reasoning_content"].as_str().map(|s| s.to_string());
+        
+        let usage = &resp_json["usage"];
+        let prompt_tokens = usage["prompt_tokens"].as_u64().unwrap_or(0) as u32;
+        let completion_tokens = usage["completion_tokens"].as_u64().unwrap_or(0) as u32;
+        let total_tokens = usage["total_tokens"].as_u64().unwrap_or(0) as u32;
+
+        Ok(ProviderResponse {
+            content,
+            reasoning_content,
+            prompt_tokens,
+            completion_tokens,
+            total_tokens,
+            model: request.model,
+        })
+    }
+
+    fn estimate_tokens(&self, request: &UnifiedRequest) -> Result<u32> {
+        Ok(crate::utils::tokens::estimate_request_tokens(&request.model, request))
+    }
+
+    fn calculate_cost(&self, model: &str, prompt_tokens: u32, completion_tokens: u32, registry: &crate::models::registry::ModelRegistry) -> f64 {
+        if let Some(metadata) = registry.find_model(model) {
+            if let Some(cost) = &metadata.cost {
+                return (prompt_tokens as f64 * cost.input / 1_000_000.0) + 
+                       (completion_tokens as f64 * cost.output / 1_000_000.0);
+            }
+        }
+        
+        let (prompt_rate, completion_rate) = self.pricing.iter()
+            .find(|p| model.contains(&p.model))
+            .map(|p| (p.prompt_tokens_per_million, p.completion_tokens_per_million))
+            .unwrap_or((5.0, 15.0));
+            
+        (prompt_tokens as f64 * prompt_rate / 1_000_000.0) + (completion_tokens as f64 * completion_rate / 1_000_000.0)
+    }
+
+    async fn chat_completion_stream(
+        &self,
+        request: UnifiedRequest,
+    ) -> Result<BoxStream<'static, Result<ProviderStreamChunk, AppError>>, AppError> {
+        let mut body = serde_json::json!({
+            "model": request.model,
+            "messages": request.messages.iter().map(|m| {
+                serde_json::json!({
+                    "role": m.role,
+                    "content": m.content.iter().map(|p| {
+                        match p {
+                            crate::models::ContentPart::Text { text } => serde_json::json!({ "type": "text", "text": text }),
+                            crate::models::ContentPart::Image(image_input) => {
+                                let (base64_data, mime_type) = futures::executor::block_on(image_input.to_base64()).unwrap_or_default();
+                                serde_json::json!({
+                                    "type": "image_url",
+                                    "image_url": { "url": format!("data:{};base64,{}", mime_type, base64_data) }
+                                })
+                            }
+                        }
+                    }).collect::<Vec<_>>()
+                })
+            }).collect::<Vec<_>>(),
+            "stream": true,
+        });
+
+        if let Some(temp) = request.temperature {
+            body["temperature"] = serde_json::json!(temp);
+        }
+        if let Some(max_tokens) = request.max_tokens {
+            body["max_tokens"] = serde_json::json!(max_tokens);
+        }
+
+        // Create eventsource stream
+        use reqwest_eventsource::{EventSource, Event};
+        let es = EventSource::new(self.client.post(format!("{}/chat/completions", self._config.base_url))
+            .header("Authorization", format!("Bearer {}", self.api_key))
+            .json(&body))
+            .map_err(|e| AppError::ProviderError(format!("Failed to create EventSource: {}", e)))?;
+        
+        let model = request.model.clone();
+        
+        let stream = async_stream::try_stream! {
+            let mut es = es;
+            while let Some(event) = es.next().await {
+                match event {
+                    Ok(Event::Message(msg)) => {
+                        if msg.data == "[DONE]" {
+                            break;
+                        }
+                        
+                        let chunk: Value = serde_json::from_str(&msg.data)
+                            .map_err(|e| AppError::ProviderError(format!("Failed to parse stream chunk: {}", e)))?;
+                        
+                        if let Some(choice) = chunk["choices"].get(0) {
+                            let delta = &choice["delta"];
+                            let content = delta["content"].as_str().unwrap_or_default().to_string();
+                            let reasoning_content = delta["reasoning_content"].as_str().map(|s| s.to_string());
+                            let finish_reason = choice["finish_reason"].as_str().map(|s| s.to_string());
+                            
+                            yield ProviderStreamChunk {
+                                content,
+                                reasoning_content,
+                                finish_reason,
+                                model: model.clone(),
+                            };
+                        }
+                    }
+                    Ok(_) => continue,
+                    Err(e) => {
+                        Err(AppError::ProviderError(format!("Stream error: {}", e)))?;
+                    }
+                }
+            }
+        };
+        
+        Ok(Box::pin(stream))
+    }
+}

src/providers/mod.rs

@@ -0,0 +1,141 @@
+use async_trait::async_trait;
+use anyhow::Result;
+use std::sync::Arc;
+use futures::stream::BoxStream;
+
+use crate::models::UnifiedRequest;
+use crate::errors::AppError;
+
+pub mod openai;
+pub mod gemini;
+pub mod deepseek;
+pub mod grok;
+pub mod ollama;
+
+#[async_trait]
+pub trait Provider: Send + Sync {
+    /// Get provider name (e.g., "openai", "gemini")
+    fn name(&self) -> &str;
+
+    /// Check if provider supports a specific model
+    fn supports_model(&self, model: &str) -> bool;
+
+    /// Check if provider supports multimodal (images, etc.)
+    fn supports_multimodal(&self) -> bool;
+
+    /// Process a chat completion request
+    async fn chat_completion(
+        &self,
+        request: UnifiedRequest,
+    ) -> Result<ProviderResponse, AppError>;
+
+    /// Process a streaming chat completion request
+    async fn chat_completion_stream(
+        &self,
+        request: UnifiedRequest,
+    ) -> Result<BoxStream<'static, Result<ProviderStreamChunk, AppError>>, AppError>;
+
+    /// Estimate token count for a request (for cost calculation)
+    fn estimate_tokens(&self, request: &UnifiedRequest) -> Result<u32>;
+
+    /// Calculate cost based on token usage and model using the registry
+    fn calculate_cost(&self, model: &str, prompt_tokens: u32, completion_tokens: u32, registry: &crate::models::registry::ModelRegistry) -> f64;
+}
+
+pub struct ProviderResponse {
+    pub content: String,
+    pub reasoning_content: Option<String>,
+    pub prompt_tokens: u32,
+    pub completion_tokens: u32,
+    pub total_tokens: u32,
+    pub model: String,
+}
+
+#[derive(Debug, Clone)]
+pub struct ProviderStreamChunk {
+    pub content: String,
+    pub reasoning_content: Option<String>,
+    pub finish_reason: Option<String>,
+    pub model: String,
+}
+
+#[derive(Clone)]
+pub struct ProviderManager {
+    providers: Vec<Arc<dyn Provider>>,
+}
+
+impl ProviderManager {
+    pub fn new() -> Self {
+        Self {
+            providers: Vec::new(),
+        }
+    }
+
+    pub fn add_provider(&mut self, provider: Arc<dyn Provider>) {
+        self.providers.push(provider);
+    }
+
+    pub fn get_provider_for_model(&self, model: &str) -> Option<Arc<dyn Provider>> {
+        self.providers.iter()
+            .find(|p| p.supports_model(model))
+            .map(|p| Arc::clone(p))
+    }
+
+    pub fn get_provider(&self, name: &str) -> Option<Arc<dyn Provider>> {
+        self.providers.iter()
+            .find(|p| p.name() == name)
+            .map(|p| Arc::clone(p))
+    }
+}
+
+// Create placeholder provider implementations
+pub mod placeholder {
+    use super::*;
+
+    pub struct PlaceholderProvider {
+        name: String,
+    }
+
+    impl PlaceholderProvider {
+        pub fn new(name: &str) -> Self {
+            Self { name: name.to_string() }
+        }
+    }
+
+    #[async_trait]
+    impl Provider for PlaceholderProvider {
+        fn name(&self) -> &str {
+            &self.name
+        }
+
+        fn supports_model(&self, _model: &str) -> bool {
+            false
+        }
+
+        fn supports_multimodal(&self) -> bool {
+            false
+        }
+
+        async fn chat_completion_stream(
+            &self,
+            _request: UnifiedRequest,
+        ) -> Result<BoxStream<'static, Result<ProviderStreamChunk, AppError>>, AppError> {
+            Err(AppError::ProviderError("Streaming not supported for placeholder provider".to_string()))
+        }
+
+        async fn chat_completion(
+            &self,
+            _request: UnifiedRequest,
+        ) -> Result<ProviderResponse, AppError> {
+            Err(AppError::ProviderError(format!("Provider {} not implemented", self.name)))
+        }
+
+        fn estimate_tokens(&self, _request: &UnifiedRequest) -> Result<u32> {
+            Ok(0)
+        }
+
+        fn calculate_cost(&self, _model: &str, _prompt_tokens: u32, _completion_tokens: u32, _registry: &crate::models::registry::ModelRegistry) -> f64 {
+            0.0
+        }
+    }
+}

src/providers/ollama.rs

@@ -0,0 +1,205 @@
+use async_trait::async_trait;
+use anyhow::Result;
+use futures::stream::{BoxStream, StreamExt};
+use serde_json::Value;
+
+use crate::{
+    models::UnifiedRequest,
+    errors::AppError,
+    config::AppConfig,
+};
+use super::{ProviderResponse, ProviderStreamChunk};
+
+pub struct OllamaProvider {
+    client: reqwest::Client,
+    _config: crate::config::OllamaConfig,
+    pricing: Vec<crate::config::ModelPricing>,
+}
+
+impl OllamaProvider {
+    pub fn new(config: &crate::config::OllamaConfig, app_config: &AppConfig) -> Result<Self> {
+        Ok(Self {
+            client: reqwest::Client::new(),
+            _config: config.clone(),
+            pricing: app_config.pricing.ollama.clone(),
+        })
+    }
+}
+
+#[async_trait]
+impl super::Provider for OllamaProvider {
+    fn name(&self) -> &str {
+        "ollama"
+    }
+
+    fn supports_model(&self, model: &str) -> bool {
+        self._config.models.iter().any(|m| m == model) || model.starts_with("ollama/")
+    }
+
+    fn supports_multimodal(&self) -> bool {
+        true
+    }
+
+    async fn chat_completion(
+        &self,
+        request: UnifiedRequest,
+    ) -> Result<ProviderResponse, AppError> {
+        let model = request.model.strip_prefix("ollama/").unwrap_or(&request.model).to_string();
+
+        let mut body = serde_json::json!({
+            "model": model,
+            "messages": request.messages.iter().map(|m| {
+                serde_json::json!({
+                    "role": m.role,
+                    "content": m.content.iter().map(|p| {
+                        match p {
+                            crate::models::ContentPart::Text { text } => serde_json::json!({ "type": "text", "text": text }),
+                            crate::models::ContentPart::Image(image_input) => {
+                                let (base64_data, mime_type) = futures::executor::block_on(image_input.to_base64()).unwrap_or_default();
+                                serde_json::json!({
+                                    "type": "image_url",
+                                    "image_url": { "url": format!("data:{};base64,{}", mime_type, base64_data) }
+                                })
+                            }
+                        }
+                    }).collect::<Vec<_>>()
+                })
+            }).collect::<Vec<_>>(),
+            "stream": false,
+        });
+
+        if let Some(temp) = request.temperature {
+            body["temperature"] = serde_json::json!(temp);
+        }
+        if let Some(max_tokens) = request.max_tokens {
+            body["max_tokens"] = serde_json::json!(max_tokens);
+        }
+
+        let response = self.client.post(format!("{}/chat/completions", self._config.base_url))
+            .json(&body)
+            .send()
+            .await
+            .map_err(|e| AppError::ProviderError(e.to_string()))?;
+
+        if !response.status().is_success() {
+            let error_text = response.text().await.unwrap_or_default();
+            return Err(AppError::ProviderError(format!("Ollama API error: {}", error_text)));
+        }
+
+        let resp_json: Value = response.json().await.map_err(|e| AppError::ProviderError(e.to_string()))?;
+        
+        let choice = resp_json["choices"].get(0).ok_or_else(|| AppError::ProviderError("No choices in response".to_string()))?;
+        let message = &choice["message"];
+        
+        let content = message["content"].as_str().unwrap_or_default().to_string();
+        let reasoning_content = message["reasoning_content"].as_str().or_else(|| message["thought"].as_str()).map(|s| s.to_string());
+        
+        let usage = &resp_json["usage"];
+        let prompt_tokens = usage["prompt_tokens"].as_u64().unwrap_or(0) as u32;
+        let completion_tokens = usage["completion_tokens"].as_u64().unwrap_or(0) as u32;
+        let total_tokens = usage["total_tokens"].as_u64().unwrap_or(0) as u32;
+
+        Ok(ProviderResponse {
+            content,
+            reasoning_content,
+            prompt_tokens,
+            completion_tokens,
+            total_tokens,
+            model: request.model,
+        })
+    }
+
+    fn estimate_tokens(&self, request: &UnifiedRequest) -> Result<u32> {
+        Ok(crate::utils::tokens::estimate_request_tokens(&request.model, request))
+    }
+
+    fn calculate_cost(&self, model: &str, prompt_tokens: u32, completion_tokens: u32, registry: &crate::models::registry::ModelRegistry) -> f64 {
+        if let Some(metadata) = registry.find_model(model) {
+            if let Some(cost) = &metadata.cost {
+                return (prompt_tokens as f64 * cost.input / 1_000_000.0) + 
+                       (completion_tokens as f64 * cost.output / 1_000_000.0);
+            }
+        }
+        
+        let (prompt_rate, completion_rate) = self.pricing.iter()
+            .find(|p| model.contains(&p.model))
+            .map(|p| (p.prompt_tokens_per_million, p.completion_tokens_per_million))
+            .unwrap_or((0.0, 0.0));
+            
+        (prompt_tokens as f64 * prompt_rate / 1_000_000.0) + (completion_tokens as f64 * completion_rate / 1_000_000.0)
+    }
+
+    async fn chat_completion_stream(
+        &self,
+        request: UnifiedRequest,
+    ) -> Result<BoxStream<'static, Result<ProviderStreamChunk, AppError>>, AppError> {
+        let model = request.model.strip_prefix("ollama/").unwrap_or(&request.model).to_string();
+
+        let mut body = serde_json::json!({
+            "model": model,
+            "messages": request.messages.iter().map(|m| {
+                serde_json::json!({
+                    "role": m.role,
+                    "content": m.content.iter().map(|p| {
+                        match p {
+                            crate::models::ContentPart::Text { text } => serde_json::json!({ "type": "text", "text": text }),
+                            crate::models::ContentPart::Image(_) => serde_json::json!({ "type": "text", "text": "[Image]" }),
+                        }
+                    }).collect::<Vec<_>>()
+                })
+            }).collect::<Vec<_>>(),
+            "stream": true,
+        });
+
+        if let Some(temp) = request.temperature {
+            body["temperature"] = serde_json::json!(temp);
+        }
+        if let Some(max_tokens) = request.max_tokens {
+            body["max_tokens"] = serde_json::json!(max_tokens);
+        }
+
+        // Create eventsource stream
+        use reqwest_eventsource::{EventSource, Event};
+        let es = EventSource::new(self.client.post(format!("{}/chat/completions", self._config.base_url))
+            .json(&body))
+            .map_err(|e| AppError::ProviderError(format!("Failed to create EventSource: {}", e)))?;
+        
+        let model_name = request.model.clone();
+        
+        let stream = async_stream::try_stream! {
+            let mut es = es;
+            while let Some(event) = es.next().await {
+                match event {
+                    Ok(Event::Message(msg)) => {
+                        if msg.data == "[DONE]" {
+                            break;
+                        }
+                        
+                        let chunk: Value = serde_json::from_str(&msg.data)
+                            .map_err(|e| AppError::ProviderError(format!("Failed to parse stream chunk: {}", e)))?;
+                        
+                        if let Some(choice) = chunk["choices"].get(0) {
+                            let delta = &choice["delta"];
+                            let content = delta["content"].as_str().unwrap_or_default().to_string();
+                            let reasoning_content = delta["reasoning_content"].as_str().or_else(|| delta["thought"].as_str()).map(|s| s.to_string());
+                            let finish_reason = choice["finish_reason"].as_str().map(|s| s.to_string());
+                            
+                            yield ProviderStreamChunk {
+                                content,
+                                reasoning_content,
+                                finish_reason,
+                                model: model_name.clone(),
+                            };
+                        }
+                    }
+                    Ok(_) => continue,
+                    Err(e) => {
+                        Err(AppError::ProviderError(format!("Stream error: {}", e)))?;
+                    }
+                }
+            }
+        };
+        
+        Ok(Box::pin(stream))
+    }
+}

src/providers/openai.rs

@@ -0,0 +1,213 @@
+use async_trait::async_trait;
+use anyhow::Result;
+use futures::stream::{BoxStream, StreamExt};
+use serde_json::Value;
+
+use crate::{
+    models::UnifiedRequest,
+    errors::AppError,
+    config::AppConfig,
+};
+use super::{ProviderResponse, ProviderStreamChunk};
+
+pub struct OpenAIProvider {
+    client: reqwest::Client,
+    _config: crate::config::OpenAIConfig,
+    api_key: String,
+    pricing: Vec<crate::config::ModelPricing>,
+}
+
+impl OpenAIProvider {
+    pub fn new(config: &crate::config::OpenAIConfig, app_config: &AppConfig) -> Result<Self> {
+        let api_key = app_config.get_api_key("openai")?;
+        
+        Ok(Self {
+            client: reqwest::Client::new(),
+            _config: config.clone(),
+            api_key,
+            pricing: app_config.pricing.openai.clone(),
+        })
+    }
+}
+
+#[async_trait]
+impl super::Provider for OpenAIProvider {
+    fn name(&self) -> &str {
+        "openai"
+    }
+
+    fn supports_model(&self, model: &str) -> bool {
+        model.starts_with("gpt-") || model.starts_with("o1-") || model.starts_with("o3-")
+    }
+
+    fn supports_multimodal(&self) -> bool {
+        true
+    }
+
+    async fn chat_completion(
+        &self,
+        request: UnifiedRequest,
+    ) -> Result<ProviderResponse, AppError> {
+        let mut body = serde_json::json!({
+            "model": request.model,
+            "messages": request.messages.iter().map(|m| {
+                serde_json::json!({
+                    "role": m.role,
+                    "content": m.content.iter().map(|p| {
+                        match p {
+                            crate::models::ContentPart::Text { text } => serde_json::json!({ "type": "text", "text": text }),
+                            crate::models::ContentPart::Image(image_input) => {
+                                let (base64_data, mime_type) = futures::executor::block_on(image_input.to_base64()).unwrap_or_default();
+                                serde_json::json!({
+                                    "type": "image_url",
+                                    "image_url": { "url": format!("data:{};base64,{}", mime_type, base64_data) }
+                                })
+                            }
+                        }
+                    }).collect::<Vec<_>>()
+                })
+            }).collect::<Vec<_>>(),
+            "stream": false,
+        });
+
+        if let Some(temp) = request.temperature {
+            body["temperature"] = serde_json::json!(temp);
+        }
+        if let Some(max_tokens) = request.max_tokens {
+            body["max_tokens"] = serde_json::json!(max_tokens);
+        }
+
+        let response = self.client.post(format!("{}/chat/completions", self._config.base_url))
+            .header("Authorization", format!("Bearer {}", self.api_key))
+            .json(&body)
+            .send()
+            .await
+            .map_err(|e| AppError::ProviderError(e.to_string()))?;
+
+        if !response.status().is_success() {
+            let error_text = response.text().await.unwrap_or_default();
+            return Err(AppError::ProviderError(format!("OpenAI API error: {}", error_text)));
+        }
+
+        let resp_json: Value = response.json().await.map_err(|e| AppError::ProviderError(e.to_string()))?;
+        
+        let choice = resp_json["choices"].get(0).ok_or_else(|| AppError::ProviderError("No choices in response".to_string()))?;
+        let message = &choice["message"];
+        
+        let content = message["content"].as_str().unwrap_or_default().to_string();
+        let reasoning_content = message["reasoning_content"].as_str().map(|s| s.to_string());
+        
+        let usage = &resp_json["usage"];
+        let prompt_tokens = usage["prompt_tokens"].as_u64().unwrap_or(0) as u32;
+        let completion_tokens = usage["completion_tokens"].as_u64().unwrap_or(0) as u32;
+        let total_tokens = usage["total_tokens"].as_u64().unwrap_or(0) as u32;
+
+        Ok(ProviderResponse {
+            content,
+            reasoning_content,
+            prompt_tokens,
+            completion_tokens,
+            total_tokens,
+            model: request.model,
+        })
+    }
+
+    fn estimate_tokens(&self, request: &UnifiedRequest) -> Result<u32> {
+        Ok(crate::utils::tokens::estimate_request_tokens(&request.model, request))
+    }
+
+    fn calculate_cost(&self, model: &str, prompt_tokens: u32, completion_tokens: u32, registry: &crate::models::registry::ModelRegistry) -> f64 {
+        if let Some(metadata) = registry.find_model(model) {
+            if let Some(cost) = &metadata.cost {
+                return (prompt_tokens as f64 * cost.input / 1_000_000.0) + 
+                       (completion_tokens as f64 * cost.output / 1_000_000.0);
+            }
+        }
+        
+        let (prompt_rate, completion_rate) = self.pricing.iter()
+            .find(|p| model.contains(&p.model))
+            .map(|p| (p.prompt_tokens_per_million, p.completion_tokens_per_million))
+            .unwrap_or((0.15, 0.60));
+            
+        (prompt_tokens as f64 * prompt_rate / 1_000_000.0) + (completion_tokens as f64 * completion_rate / 1_000_000.0)
+    }
+
+    async fn chat_completion_stream(
+        &self,
+        request: UnifiedRequest,
+    ) -> Result<BoxStream<'static, Result<ProviderStreamChunk, AppError>>, AppError> {
+        let mut body = serde_json::json!({
+            "model": request.model,
+            "messages": request.messages.iter().map(|m| {
+                serde_json::json!({
+                    "role": m.role,
+                    "content": m.content.iter().map(|p| {
+                        match p {
+                            crate::models::ContentPart::Text { text } => serde_json::json!({ "type": "text", "text": text }),
+                            crate::models::ContentPart::Image(image_input) => {
+                                let (base64_data, mime_type) = futures::executor::block_on(image_input.to_base64()).unwrap_or_default();
+                                serde_json::json!({
+                                    "type": "image_url",
+                                    "image_url": { "url": format!("data:{};base64,{}", mime_type, base64_data) }
+                                })
+                            }
+                        }
+                    }).collect::<Vec<_>>()
+                })
+            }).collect::<Vec<_>>(),
+            "stream": true,
+        });
+
+        if let Some(temp) = request.temperature {
+            body["temperature"] = serde_json::json!(temp);
+        }
+        if let Some(max_tokens) = request.max_tokens {
+            body["max_tokens"] = serde_json::json!(max_tokens);
+        }
+
+        // Create eventsource stream
+        use reqwest_eventsource::{EventSource, Event};
+        let es = EventSource::new(self.client.post(format!("{}/chat/completions", self._config.base_url))
+            .header("Authorization", format!("Bearer {}", self.api_key))
+            .json(&body))
+            .map_err(|e| AppError::ProviderError(format!("Failed to create EventSource: {}", e)))?;
+        
+        let model = request.model.clone();
+        
+        let stream = async_stream::try_stream! {
+            let mut es = es;
+            while let Some(event) = es.next().await {
+                match event {
+                    Ok(Event::Message(msg)) => {
+                        if msg.data == "[DONE]" {
+                            break;
+                        }
+                        
+                        let chunk: Value = serde_json::from_str(&msg.data)
+                            .map_err(|e| AppError::ProviderError(format!("Failed to parse stream chunk: {}", e)))?;
+                        
+                        if let Some(choice) = chunk["choices"].get(0) {
+                            let delta = &choice["delta"];
+                            let content = delta["content"].as_str().unwrap_or_default().to_string();
+                            let reasoning_content = delta["reasoning_content"].as_str().map(|s| s.to_string());
+                            let finish_reason = choice["finish_reason"].as_str().map(|s| s.to_string());
+                            
+                            yield ProviderStreamChunk {
+                                content,
+                                reasoning_content,
+                                finish_reason,
+                                model: model.clone(),
+                            };
+                        }
+                    }
+                    Ok(_) => continue,
+                    Err(e) => {
+                        Err(AppError::ProviderError(format!("Stream error: {}", e)))?;
+                    }
+                }
+            }
+        };
+        
+        Ok(Box::pin(stream))
+    }
+}

src/rate_limiting/mod.rs

@@ -0,0 +1,359 @@
+//! Rate limiting and circuit breaking for LLM proxy
+//!
+//! This module provides:
+//! 1. Per-client rate limiting using governor crate
+//! 2. Provider circuit breaking to handle API failures
+//! 3. Global rate limiting for overall system protection
+
+use std::sync::Arc;
+use std::collections::HashMap;
+use std::time::Instant;
+use tokio::sync::RwLock;
+use tracing::{info, warn};
+use anyhow::Result;
+
+/// Rate limiter configuration
+#[derive(Debug, Clone)]
+pub struct RateLimiterConfig {
+    /// Requests per minute per client
+    pub requests_per_minute: u32,
+    /// Burst size (maximum burst capacity)
+    pub burst_size: u32,
+    /// Global requests per minute (across all clients)
+    pub global_requests_per_minute: u32,
+}
+
+impl Default for RateLimiterConfig {
+    fn default() -> Self {
+        Self {
+            requests_per_minute: 60,  // 1 request per second per client
+            burst_size: 10,           // Allow bursts of up to 10 requests
+            global_requests_per_minute: 600, // 10 requests per second globally
+        }
+    }
+}
+
+/// Circuit breaker state
+#[derive(Debug, Clone, Copy, PartialEq)]
+pub enum CircuitState {
+    Closed,      // Normal operation
+    Open,        // Circuit is open, requests fail fast
+    HalfOpen,    // Testing if service has recovered
+}
+
+/// Circuit breaker configuration
+#[derive(Debug, Clone)]
+pub struct CircuitBreakerConfig {
+    /// Failure threshold to open circuit
+    pub failure_threshold: u32,
+    /// Time window for failure counting (seconds)
+    pub failure_window_secs: u64,
+    /// Time to wait before trying half-open state (seconds)
+    pub reset_timeout_secs: u64,
+    /// Success threshold to close circuit
+    pub success_threshold: u32,
+}
+
+impl Default for CircuitBreakerConfig {
+    fn default() -> Self {
+        Self {
+            failure_threshold: 5,      // 5 failures
+            failure_window_secs: 60,   // within 60 seconds
+            reset_timeout_secs: 30,    // wait 30 seconds before half-open
+            success_threshold: 3,      // 3 successes to close circuit
+        }
+    }
+}
+
+/// Simple token bucket rate limiter for a single client
+#[derive(Debug)]
+struct TokenBucket {
+    tokens: f64,
+    capacity: f64,
+    refill_rate: f64, // tokens per second
+    last_refill: Instant,
+}
+
+impl TokenBucket {
+    fn new(capacity: f64, refill_rate: f64) -> Self {
+        Self {
+            tokens: capacity,
+            capacity,
+            refill_rate,
+            last_refill: Instant::now(),
+        }
+    }
+
+    fn refill(&mut self) {
+        let now = Instant::now();
+        let elapsed = now.duration_since(self.last_refill).as_secs_f64();
+        let new_tokens = elapsed * self.refill_rate;
+        
+        self.tokens = (self.tokens + new_tokens).min(self.capacity);
+        self.last_refill = now;
+    }
+
+    fn try_acquire(&mut self, tokens: f64) -> bool {
+        self.refill();
+        
+        if self.tokens >= tokens {
+            self.tokens -= tokens;
+            true
+        } else {
+            false
+        }
+    }
+}
+
+/// Circuit breaker for a provider
+#[derive(Debug)]
+pub struct ProviderCircuitBreaker {
+    state: CircuitState,
+    failure_count: u32,
+    success_count: u32,
+    last_failure_time: Option<std::time::Instant>,
+    last_state_change: std::time::Instant,
+    config: CircuitBreakerConfig,
+}
+
+impl ProviderCircuitBreaker {
+    pub fn new(config: CircuitBreakerConfig) -> Self {
+        Self {
+            state: CircuitState::Closed,
+            failure_count: 0,
+            success_count: 0,
+            last_failure_time: None,
+            last_state_change: std::time::Instant::now(),
+            config,
+        }
+    }
+
+    /// Check if request is allowed
+    pub fn allow_request(&mut self) -> bool {
+        match self.state {
+            CircuitState::Closed => true,
+            CircuitState::Open => {
+                // Check if reset timeout has passed
+                let elapsed = self.last_state_change.elapsed();
+                if elapsed.as_secs() >= self.config.reset_timeout_secs {
+                    self.state = CircuitState::HalfOpen;
+                    self.last_state_change = std::time::Instant::now();
+                    info!("Circuit breaker transitioning to half-open state");
+                    true
+                } else {
+                    false
+                }
+            }
+            CircuitState::HalfOpen => true,
+        }
+    }
+
+    /// Record a successful request
+    pub fn record_success(&mut self) {
+        match self.state {
+            CircuitState::Closed => {
+                // Reset failure count on success
+                self.failure_count = 0;
+                self.last_failure_time = None;
+            }
+            CircuitState::HalfOpen => {
+                self.success_count += 1;
+                if self.success_count >= self.config.success_threshold {
+                    self.state = CircuitState::Closed;
+                    self.success_count = 0;
+                    self.failure_count = 0;
+                    self.last_state_change = std::time::Instant::now();
+                    info!("Circuit breaker closed after successful requests");
+                }
+            }
+            CircuitState::Open => {
+                // Should not happen, but handle gracefully
+            }
+        }
+    }
+
+    /// Record a failed request
+    pub fn record_failure(&mut self) {
+        let now = std::time::Instant::now();
+        
+        // Check if failure window has expired
+        if let Some(last_failure) = self.last_failure_time {
+            if now.duration_since(last_failure).as_secs() > self.config.failure_window_secs {
+                // Reset failure count if window expired
+                self.failure_count = 0;
+            }
+        }
+        
+        self.failure_count += 1;
+        self.last_failure_time = Some(now);
+        
+        if self.failure_count >= self.config.failure_threshold && self.state == CircuitState::Closed {
+            self.state = CircuitState::Open;
+            self.last_state_change = now;
+            warn!("Circuit breaker opened due to {} failures", self.failure_count);
+        } else if self.state == CircuitState::HalfOpen {
+            // Failure in half-open state, go back to open
+            self.state = CircuitState::Open;
+            self.success_count = 0;
+            self.last_state_change = now;
+            warn!("Circuit breaker re-opened after failure in half-open state");
+        }
+    }
+
+    /// Get current state
+    pub fn state(&self) -> CircuitState {
+        self.state
+    }
+}
+
+/// Rate limiting and circuit breaking manager
+#[derive(Debug)]
+pub struct RateLimitManager {
+    client_buckets: Arc<RwLock<HashMap<String, TokenBucket>>>,
+    global_bucket: Arc<RwLock<TokenBucket>>,
+    circuit_breakers: Arc<RwLock<HashMap<String, ProviderCircuitBreaker>>>,
+    config: RateLimiterConfig,
+    circuit_config: CircuitBreakerConfig,
+}
+
+impl RateLimitManager {
+    pub fn new(config: RateLimiterConfig, circuit_config: CircuitBreakerConfig) -> Self {
+        // Convert requests per minute to tokens per second
+        let global_refill_rate = config.global_requests_per_minute as f64 / 60.0;
+        
+        Self {
+            client_buckets: Arc::new(RwLock::new(HashMap::new())),
+            global_bucket: Arc::new(RwLock::new(TokenBucket::new(
+                config.burst_size as f64,
+                global_refill_rate,
+            ))),
+            circuit_breakers: Arc::new(RwLock::new(HashMap::new())),
+            config,
+            circuit_config,
+        }
+    }
+
+    /// Check if a client request is allowed
+    pub async fn check_client_request(&self, client_id: &str) -> Result<bool> {
+        // Check global rate limit first (1 token per request)
+        {
+            let mut global_bucket = self.global_bucket.write().await;
+            if !global_bucket.try_acquire(1.0) {
+                warn!("Global rate limit exceeded");
+                return Ok(false);
+            }
+        }
+        
+        // Check client-specific rate limit
+        let mut buckets = self.client_buckets.write().await;
+        let bucket = buckets
+            .entry(client_id.to_string())
+            .or_insert_with(|| {
+                TokenBucket::new(
+                    self.config.burst_size as f64,
+                    self.config.requests_per_minute as f64 / 60.0,
+                )
+            });
+        
+        Ok(bucket.try_acquire(1.0))
+    }
+
+    /// Check if provider requests are allowed (circuit breaker)
+    pub async fn check_provider_request(&self, provider_name: &str) -> Result<bool> {
+        let mut breakers = self.circuit_breakers.write().await;
+        let breaker = breakers
+            .entry(provider_name.to_string())
+            .or_insert_with(|| ProviderCircuitBreaker::new(self.circuit_config.clone()));
+        
+        Ok(breaker.allow_request())
+    }
+
+    /// Record provider success
+    pub async fn record_provider_success(&self, provider_name: &str) {
+        let mut breakers = self.circuit_breakers.write().await;
+        if let Some(breaker) = breakers.get_mut(provider_name) {
+            breaker.record_success();
+        }
+    }
+
+    /// Record provider failure
+    pub async fn record_provider_failure(&self, provider_name: &str) {
+        let mut breakers = self.circuit_breakers.write().await;
+        let breaker = breakers
+            .entry(provider_name.to_string())
+            .or_insert_with(|| ProviderCircuitBreaker::new(self.circuit_config.clone()));
+        
+        breaker.record_failure();
+    }
+
+    /// Get provider circuit state
+    pub async fn get_provider_state(&self, provider_name: &str) -> CircuitState {
+        let breakers = self.circuit_breakers.read().await;
+        breakers
+            .get(provider_name)
+            .map(|b| b.state())
+            .unwrap_or(CircuitState::Closed)
+    }
+}
+
+/// Axum middleware for rate limiting
+pub mod middleware {
+    use super::*;
+    use axum::{
+        extract::{Request, State},
+        middleware::Next,
+        response::Response,
+    };
+    use crate::errors::AppError;
+    use crate::state::AppState;
+
+
+    /// Rate limiting middleware
+    pub async fn rate_limit_middleware(
+        State(state): State<AppState>,
+        request: Request,
+        next: Next,
+    ) -> Result<Response, AppError> {
+        // Extract client ID from authentication header
+        let client_id = extract_client_id_from_request(&request);
+
+        // Check rate limits
+        if !state.rate_limit_manager.check_client_request(&client_id).await? {
+            return Err(AppError::RateLimitError(
+                "Rate limit exceeded".to_string()
+            ));
+        }
+
+        Ok(next.run(request).await)
+    }
+    
+    /// Extract client ID from request (helper function)
+    fn extract_client_id_from_request(request: &Request) -> String {
+        // Try to extract from Authorization header
+        if let Some(auth_header) = request.headers().get("Authorization") {
+            if let Ok(auth_str) = auth_header.to_str() {
+                if auth_str.starts_with("Bearer ") {
+                    let token = &auth_str[7..];
+                    // Use token hash as client ID (same logic as auth module)
+                    return format!("client_{}", &token[..8.min(token.len())]);
+                }
+            }
+        }
+        
+        // Fallback to anonymous
+        "anonymous".to_string()
+    }
+
+    /// Circuit breaker middleware for provider requests
+    pub async fn circuit_breaker_middleware(
+        provider_name: &str,
+        state: &AppState,
+    ) -> Result<(), AppError> {
+        if !state.rate_limit_manager.check_provider_request(provider_name).await? {
+            return Err(AppError::ProviderError(
+                format!("Provider {} is currently unavailable (circuit breaker open)", provider_name)
+            ));
+        }
+        Ok(())
+    }
+}

src/server/mod.rs

@@ -0,0 +1,224 @@
+use uuid::Uuid;
+use axum::{
+    extract::State,
+    routing::post,
+    Json, Router,
+    response::sse::{Event, Sse},
+    response::IntoResponse,
+};
+use futures::stream::StreamExt;
+use tracing::{info, warn};
+
+use crate::{
+    auth::AuthenticatedClient,
+    errors::AppError,
+    models::{ChatCompletionRequest, ChatCompletionResponse, ChatCompletionStreamResponse, ChatStreamChoice, ChatStreamDelta, ChatMessage, ChatChoice, Usage},
+    state::AppState,
+    rate_limiting,
+};
+
+pub fn router(state: AppState) -> Router {
+    Router::new()
+        .route("/v1/chat/completions", post(chat_completions))
+        .layer(axum::middleware::from_fn_with_state(
+            state.clone(),
+            rate_limiting::middleware::rate_limit_middleware,
+        ))
+        .with_state(state)
+}
+
+async fn chat_completions(
+    State(state): State<AppState>,
+    auth: AuthenticatedClient,
+    Json(request): Json<ChatCompletionRequest>,
+) -> Result<axum::response::Response, AppError> {
+    let start_time = std::time::Instant::now();
+    let client_id = auth.client_id.clone();
+    let model = request.model.clone();
+    
+    info!("Chat completion request from client {} for model {}", client_id, model);
+    
+    // Find appropriate provider for the model
+    let provider = state.provider_manager.get_provider_for_model(&request.model)
+        .ok_or_else(|| AppError::ProviderError(format!("No provider found for model: {}", request.model)))?;
+    
+    let provider_name = provider.name().to_string();
+    
+    // Check circuit breaker for this provider
+    rate_limiting::middleware::circuit_breaker_middleware(&provider_name, &state).await?;
+
+    // Convert to unified request format
+    let mut unified_request = crate::models::UnifiedRequest::try_from(request)
+        .map_err(|e| AppError::ValidationError(e.to_string()))?;
+    
+    // Set client_id from authentication
+    unified_request.client_id = client_id.clone();
+
+    // Hydrate images if present
+    if unified_request.has_images {
+        unified_request.hydrate_images().await
+            .map_err(|e| AppError::ValidationError(format!("Failed to process images: {}", e)))?;
+    }
+
+    // Check if streaming is requested
+    if unified_request.stream {
+        // Estimate prompt tokens for logging later
+        let prompt_tokens = crate::utils::tokens::estimate_request_tokens(&model, &unified_request);
+        let has_images = unified_request.has_images;
+
+        // Handle streaming response
+        let stream_result = provider.chat_completion_stream(unified_request).await;
+        
+        match stream_result {
+            Ok(stream) => {
+                // Record provider success
+                state.rate_limit_manager.record_provider_success(&provider_name).await;
+                
+                // Wrap with AggregatingStream for token counting and database logging
+                let aggregating_stream = crate::utils::streaming::AggregatingStream::new(
+                    stream,
+                    client_id.clone(),
+                    provider.clone(),
+                    model.clone(),
+                    prompt_tokens,
+                    has_images,
+                    state.request_logger.clone(),
+                    state.client_manager.clone(),
+                    state.model_registry.clone(),
+                );
+
+                // Create SSE stream from aggregating stream
+                let sse_stream = aggregating_stream.map(move |chunk_result| {
+                    match chunk_result {
+                        Ok(chunk) => {
+                            // Convert provider chunk to OpenAI-compatible SSE event
+                            let response = ChatCompletionStreamResponse {
+                                id: format!("chatcmpl-{}", Uuid::new_v4()),
+                                object: "chat.completion.chunk".to_string(),
+                                created: chrono::Utc::now().timestamp() as u64,
+                                model: chunk.model.clone(),
+                                choices: vec![ChatStreamChoice {
+                                    index: 0,
+                                    delta: ChatStreamDelta {
+                                        role: None,
+                                        content: Some(chunk.content),
+                                        reasoning_content: chunk.reasoning_content,
+                                    },
+                                    finish_reason: chunk.finish_reason,
+                                }],
+                            };
+                            
+                            Ok(Event::default().json_data(response).unwrap())
+                        }
+                        Err(e) => {
+                            warn!("Error in streaming response: {}", e);
+                            Err(e)
+                        }
+                    }
+                });
+                
+                Ok(Sse::new(sse_stream).into_response())
+            }
+            Err(e) => {
+                // Record provider failure
+                state.rate_limit_manager.record_provider_failure(&provider_name).await;
+                
+                // Log failed request
+                let duration = start_time.elapsed();
+                warn!("Streaming request failed after {:?}: {}", duration, e);
+                
+                Err(e)
+            }
+        }
+    } else {
+        // Handle non-streaming response
+        let result = provider.chat_completion(unified_request).await;
+
+        match result {
+            Ok(response) => {
+                // Record provider success
+                state.rate_limit_manager.record_provider_success(&provider_name).await;
+                
+                let duration = start_time.elapsed();
+                let cost = provider.calculate_cost(&response.model, response.prompt_tokens, response.completion_tokens, &state.model_registry);
+
+                // Log request to database
+                state.request_logger.log_request(crate::logging::RequestLog {
+                    timestamp: chrono::Utc::now(),
+                    client_id: client_id.clone(),
+                    provider: provider_name.clone(),
+                    model: response.model.clone(),
+                    prompt_tokens: response.prompt_tokens,
+                    completion_tokens: response.completion_tokens,
+                    total_tokens: response.total_tokens,
+                    cost,
+                    has_images: false, // TODO: check images
+                    status: "success".to_string(),
+                    error_message: None,
+                    duration_ms: duration.as_millis() as u64,
+                });
+
+                // Update client usage
+                let _ = state.client_manager.update_client_usage(
+                    &client_id,
+                    response.total_tokens as i64,
+                    cost,
+                ).await;
+
+                // Convert ProviderResponse to ChatCompletionResponse
+                let chat_response = ChatCompletionResponse {
+                    id: format!("chatcmpl-{}", Uuid::new_v4()),
+                    object: "chat.completion".to_string(),
+                    created: chrono::Utc::now().timestamp() as u64,
+                    model: response.model,
+                    choices: vec![ChatChoice {
+                        index: 0,
+                        message: ChatMessage {
+                            role: "assistant".to_string(),
+                            content: crate::models::MessageContent::Text { 
+                                content: response.content 
+                            },
+                            reasoning_content: response.reasoning_content,
+                        },
+                        finish_reason: Some("stop".to_string()),
+                    }],
+                    usage: Some(Usage {
+                        prompt_tokens: response.prompt_tokens,
+                        completion_tokens: response.completion_tokens,
+                        total_tokens: response.total_tokens,
+                    }),
+                };
+                
+                // Log successful request
+                info!("Request completed successfully in {:?}", duration);
+                
+                Ok(Json(chat_response).into_response())
+            }
+            Err(e) => {
+                // Record provider failure
+                state.rate_limit_manager.record_provider_failure(&provider_name).await;
+                
+                // Log failed request to database
+                let duration = start_time.elapsed();
+                state.request_logger.log_request(crate::logging::RequestLog {
+                    timestamp: chrono::Utc::now(),
+                    client_id: client_id.clone(),
+                    provider: provider_name.clone(),
+                    model: model.clone(),
+                    prompt_tokens: 0,
+                    completion_tokens: 0,
+                    total_tokens: 0,
+                    cost: 0.0,
+                    has_images: false,
+                    status: "error".to_string(),
+                    error_message: Some(e.to_string()),
+                    duration_ms: duration.as_millis() as u64,
+                });
+
+                warn!("Request failed after {:?}: {}", duration, e);
+                
+                Err(e)
+            }
+        }
+    }
+}

src/state/mod.rs

@@ -0,0 +1,43 @@
+use std::sync::Arc;
+use tokio::sync::broadcast;
+
+use crate::{
+    client::ClientManager, database::DbPool, providers::ProviderManager,
+    rate_limiting::RateLimitManager, logging::RequestLogger,
+    models::registry::ModelRegistry,
+};
+
+/// Shared application state
+#[derive(Clone)]
+pub struct AppState {
+    pub provider_manager: ProviderManager,
+    pub db_pool: DbPool,
+    pub rate_limit_manager: Arc<RateLimitManager>,
+    pub client_manager: Arc<ClientManager>,
+    pub request_logger: Arc<RequestLogger>,
+    pub model_registry: Arc<ModelRegistry>,
+    pub dashboard_tx: broadcast::Sender<serde_json::Value>,
+}
+
+impl AppState {
+    pub fn new(
+        provider_manager: ProviderManager,
+        db_pool: DbPool,
+        rate_limit_manager: RateLimitManager,
+        model_registry: ModelRegistry,
+    ) -> Self {
+        let client_manager = Arc::new(ClientManager::new(db_pool.clone()));
+        let (dashboard_tx, _) = broadcast::channel(100);
+        let request_logger = Arc::new(RequestLogger::new(db_pool.clone(), dashboard_tx.clone()));
+
+        Self {
+            provider_manager,
+            db_pool,
+            rate_limit_manager: Arc::new(rate_limit_manager),
+            client_manager,
+            request_logger,
+            model_registry: Arc::new(model_registry),
+            dashboard_tx,
+        }
+    }
+}

src/utils/mod.rs

@@ -0,0 +1,3 @@
+pub mod tokens;
+pub mod registry;
+pub mod streaming;

src/utils/registry.rs

@@ -0,0 +1,24 @@
+use anyhow::Result;
+use tracing::info;
+use crate::models::registry::ModelRegistry;
+
+const MODELS_DEV_URL: &str = "https://models.dev/api.json";
+
+pub async fn fetch_registry() -> Result<ModelRegistry> {
+    info!("Fetching model registry from {}", MODELS_DEV_URL);
+    
+    let client = reqwest::Client::builder()
+        .timeout(std::time::Duration::from_secs(10))
+        .build()?;
+        
+    let response = client.get(MODELS_DEV_URL).send().await?;
+    
+    if !response.status().is_success() {
+        return Err(anyhow::anyhow!("Failed to fetch registry: HTTP {}", response.status()));
+    }
+    
+    let registry: ModelRegistry = response.json().await?;
+    info!("Successfully loaded model registry");
+    
+    Ok(registry)
+}

src/utils/streaming.rs

@@ -0,0 +1,200 @@
+use futures::stream::Stream;
+use std::pin::Pin;
+use std::task::{Context, Poll};
+use std::sync::Arc;
+use crate::logging::{RequestLogger, RequestLog};
+use crate::client::ClientManager;
+use crate::providers::{Provider, ProviderStreamChunk};
+use crate::errors::AppError;
+use crate::utils::tokens::estimate_completion_tokens;
+
+pub struct AggregatingStream<S> {
+    inner: S,
+    client_id: String,
+    provider: Arc<dyn Provider>,
+    model: String,
+    prompt_tokens: u32,
+    has_images: bool,
+    accumulated_content: String,
+    accumulated_reasoning: String,
+    logger: Arc<RequestLogger>,
+    client_manager: Arc<ClientManager>,
+    model_registry: Arc<crate::models::registry::ModelRegistry>,
+    start_time: std::time::Instant,
+    has_logged: bool,
+}
+
+impl<S> AggregatingStream<S> 
+where 
+    S: Stream<Item = Result<ProviderStreamChunk, AppError>> + Unpin
+{
+    pub fn new(
+        inner: S,
+        client_id: String,
+        provider: Arc<dyn Provider>,
+        model: String,
+        prompt_tokens: u32,
+        has_images: bool,
+        logger: Arc<RequestLogger>,
+        client_manager: Arc<ClientManager>,
+        model_registry: Arc<crate::models::registry::ModelRegistry>,
+    ) -> Self {
+        Self {
+            inner,
+            client_id,
+            provider,
+            model,
+            prompt_tokens,
+            has_images,
+            accumulated_content: String::new(),
+            accumulated_reasoning: String::new(),
+            logger,
+            client_manager,
+            model_registry,
+            start_time: std::time::Instant::now(),
+            has_logged: false,
+        }
+    }
+
+    fn finalize(&mut self) {
+        if self.has_logged {
+            return;
+        }
+        self.has_logged = true;
+
+        let duration = self.start_time.elapsed();
+        let client_id = self.client_id.clone();
+        let provider_name = self.provider.name().to_string();
+        let model = self.model.clone();
+        let logger = self.logger.clone();
+        let client_manager = self.client_manager.clone();
+        let provider = self.provider.clone();
+        let prompt_tokens = self.prompt_tokens;
+        let has_images = self.has_images;
+        let registry = self.model_registry.clone();
+        
+        // Estimate completion tokens (including reasoning if present)
+        let content_tokens = estimate_completion_tokens(&self.accumulated_content, &model);
+        let reasoning_tokens = if !self.accumulated_reasoning.is_empty() {
+            estimate_completion_tokens(&self.accumulated_reasoning, &model)
+        } else {
+            0
+        };
+        
+        let completion_tokens = content_tokens + reasoning_tokens;
+        let total_tokens = prompt_tokens + completion_tokens;
+        let cost = provider.calculate_cost(&model, prompt_tokens, completion_tokens, &registry);
+
+        // Spawn a background task to log the completion
+        tokio::spawn(async move {
+            // Log to database
+            logger.log_request(RequestLog {
+                timestamp: chrono::Utc::now(),
+                client_id: client_id.clone(),
+                provider: provider_name,
+                model,
+                prompt_tokens,
+                completion_tokens,
+                total_tokens,
+                cost,
+                has_images,
+                status: "success".to_string(),
+                error_message: None,
+                duration_ms: duration.as_millis() as u64,
+            });
+
+            // Update client usage
+            let _ = client_manager.update_client_usage(
+                &client_id,
+                total_tokens as i64,
+                cost,
+            ).await;
+        });
+    }
+}
+
+impl<S> Stream for AggregatingStream<S>
+where
+    S: Stream<Item = Result<ProviderStreamChunk, AppError>> + Unpin
+{
+    type Item = Result<ProviderStreamChunk, AppError>;
+
+    fn poll_next(mut self: Pin<&mut Self>, cx: &mut Context<'_>) -> Poll<Option<Self::Item>> {
+        let result = Pin::new(&mut self.inner).poll_next(cx);
+        
+        match &result {
+            Poll::Ready(Some(Ok(chunk))) => {
+                self.accumulated_content.push_str(&chunk.content);
+                if let Some(reasoning) = &chunk.reasoning_content {
+                    self.accumulated_reasoning.push_str(reasoning);
+                }
+            }
+            Poll::Ready(Some(Err(_))) => {
+                // If there's an error, we might still want to log what we got so far?
+                // For now, just finalize if we have content
+                if !self.accumulated_content.is_empty() {
+                    self.finalize();
+                }
+            }
+            Poll::Ready(None) => {
+                self.finalize();
+            }
+            Poll::Pending => {}
+        }
+        
+        result
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use futures::stream::{self, StreamExt};
+    use anyhow::Result;
+    
+    // Simple mock provider for testing
+    struct MockProvider;
+    #[async_trait::async_trait]
+    impl Provider for MockProvider {
+        fn name(&self) -> &str { "mock" }
+        fn supports_model(&self, _model: &str) -> bool { true }
+        fn supports_multimodal(&self) -> bool { false }
+        async fn chat_completion(&self, _req: crate::models::UnifiedRequest) -> Result<crate::providers::ProviderResponse, AppError> { unimplemented!() }
+        async fn chat_completion_stream(&self, _req: crate::models::UnifiedRequest) -> Result<futures::stream::BoxStream<'static, Result<ProviderStreamChunk, AppError>>, AppError> { unimplemented!() }
+        fn estimate_tokens(&self, _req: &crate::models::UnifiedRequest) -> Result<u32> { Ok(10) }
+        fn calculate_cost(&self, _model: &str, _p: u32, _c: u32, _r: &crate::models::registry::ModelRegistry) -> f64 { 0.05 }
+    }
+
+    #[tokio::test]
+    async fn test_aggregating_stream() {
+        let chunks = vec![
+            Ok(ProviderStreamChunk { content: "Hello".to_string(), finish_reason: None, model: "test".to_string() }),
+            Ok(ProviderStreamChunk { content: " World".to_string(), finish_reason: Some("stop".to_string()), model: "test".to_string() }),
+        ];
+        let inner_stream = stream::iter(chunks);
+        
+        let pool = sqlx::SqlitePool::connect("sqlite::memory:").await.unwrap();
+        let logger = Arc::new(RequestLogger::new(pool.clone()));
+        let client_manager = Arc::new(ClientManager::new(pool.clone()));
+        let registry = Arc::new(crate::models::registry::ModelRegistry { providers: std::collections::HashMap::new() });
+        
+        let mut agg_stream = AggregatingStream::new(
+            inner_stream,
+            "client_1".to_string(),
+            Arc::new(MockProvider),
+            "test".to_string(),
+            10,
+            false,
+            logger,
+            client_manager,
+            registry,
+        );
+        
+        while let Some(item) = agg_stream.next().await {
+            assert!(item.is_ok());
+        }
+        
+        assert_eq!(agg_stream.accumulated_content, "Hello World");
+        assert!(agg_stream.has_logged);
+    }
+}

src/utils/tokens.rs

@@ -0,0 +1,51 @@
+use tiktoken_rs::get_bpe_from_model;
+use crate::models::UnifiedRequest;
+
+/// Count tokens for a given model and text
+pub fn count_tokens(model: &str, text: &str) -> u32 {
+    // If we can't get the bpe for the model, fallback to a safe default (cl100k_base for GPT-4/o1)
+    let bpe = get_bpe_from_model(model).unwrap_or_else(|_| {
+        tiktoken_rs::cl100k_base().expect("Failed to get cl100k_base encoding")
+    });
+    
+    bpe.encode_with_special_tokens(text).len() as u32
+}
+
+/// Estimate tokens for a unified request
+pub fn estimate_request_tokens(model: &str, request: &UnifiedRequest) -> u32 {
+    let mut total_tokens = 0;
+    
+    // Base tokens per message for OpenAI (approximate)
+    let tokens_per_message = 3;
+    let _tokens_per_name = 1;
+    
+    for msg in &request.messages {
+        total_tokens += tokens_per_message;
+        
+        for part in &msg.content {
+            match part {
+                crate::models::ContentPart::Text { text } => {
+                    total_tokens += count_tokens(model, text);
+                }
+                crate::models::ContentPart::Image { .. } => {
+                    // Vision models usually have a fixed cost or calculation based on size
+                    // For now, let's use a conservative estimate of 1000 tokens
+                    total_tokens += 1000;
+                }
+            }
+        }
+        
+        // Add name tokens if we had names (we don't in UnifiedMessage yet)
+        // total_tokens += tokens_per_name;
+    }
+    
+    // Add 3 tokens for the assistant reply header
+    total_tokens += 3;
+    
+    total_tokens
+}
+
+/// Estimate tokens for completion text
+pub fn estimate_completion_tokens(text: &str, model: &str) -> u32 {
+    count_tokens(model, text)
+}

static/index.html

@@ -0,0 +1,175 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <title>LLM Proxy Gateway - Admin Dashboard</title>
+    <link rel="stylesheet" href="/css/dashboard.css">
+    <link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/6.4.0/css/all.min.css">
+    <link href="https://fonts.googleapis.com/css2?family=Inter:wght@300;400;500;600;700&display=swap" rel="stylesheet">
+    <script src="https://cdn.jsdelivr.net/npm/chart.js"></script>
+    <script src="https://cdn.jsdelivr.net/npm/luxon@3.4.4/build/global/luxon.min.js"></script>
+</head>
+<body>
+    <!-- Login Screen -->
+    <div id="login-screen" class="login-container">
+        <div class="login-card">
+            <div class="login-header">
+                <i class="fas fa-robot login-icon"></i>
+                <h1>LLM Proxy Gateway</h1>
+                <p class="login-subtitle">Admin Dashboard</p>
+            </div>
+            <form id="login-form" class="login-form">
+                <div class="form-group">
+                    <label for="username">
+                        <i class="fas fa-user"></i> Username
+                    </label>
+                    <input type="text" id="username" name="username" placeholder="admin" required>
+                </div>
+                <div class="form-group">
+                    <label for="password">
+                        <i class="fas fa-lock"></i> Password
+                    </label>
+                    <input type="password" id="password" name="password" placeholder="••••••••" required>
+                </div>
+                <div class="form-group">
+                    <button type="submit" class="login-btn">
+                        <i class="fas fa-sign-in-alt"></i> Sign In
+                    </button>
+                </div>
+                <div class="login-footer">
+                    <p>Default credentials: admin / admin123</p>
+                </div>
+            </form>
+            <div id="login-error" class="error-message" style="display: none;">
+                <i class="fas fa-exclamation-circle"></i>
+                <span>Invalid credentials. Please try again.</span>
+            </div>
+        </div>
+    </div>
+
+    <!-- Main Dashboard -->
+    <div id="dashboard" class="dashboard-container" style="display: none;">
+        <!-- Sidebar -->
+        <nav class="sidebar">
+            <div class="sidebar-header">
+                <div class="logo">
+                    <i class="fas fa-robot"></i>
+                    <span>LLM Proxy</span>
+                </div>
+                <button class="sidebar-toggle" id="sidebar-toggle">
+                    <i class="fas fa-bars"></i>
+                </button>
+            </div>
+            
+            <div class="sidebar-menu">
+                <div class="menu-section">
+                    <h3 class="menu-title">MAIN</h3>
+                    <a href="#overview" class="menu-item active" data-page="overview">
+                        <i class="fas fa-tachometer-alt"></i>
+                        <span>Overview</span>
+                    </a>
+                    <a href="#analytics" class="menu-item" data-page="analytics">
+                        <i class="fas fa-chart-line"></i>
+                        <span>Analytics</span>
+                    </a>
+                    <a href="#costs" class="menu-item" data-page="costs">
+                        <i class="fas fa-dollar-sign"></i>
+                        <span>Cost Management</span>
+                    </a>
+                </div>
+                
+                <div class="menu-section">
+                    <h3 class="menu-title">MANAGEMENT</h3>
+                    <a href="#clients" class="menu-item" data-page="clients">
+                        <i class="fas fa-users"></i>
+                        <span>Client Management</span>
+                    </a>
+                    <a href="#providers" class="menu-item" data-page="providers">
+                        <i class="fas fa-server"></i>
+                        <span>Providers</span>
+                    </a>
+                    <a href="#monitoring" class="menu-item" data-page="monitoring">
+                        <i class="fas fa-heartbeat"></i>
+                        <span>Real-time Monitoring</span>
+                    </a>
+                </div>
+                
+                <div class="menu-section">
+                    <h3 class="menu-title">SYSTEM</h3>
+                    <a href="#settings" class="menu-item" data-page="settings">
+                        <i class="fas fa-cog"></i>
+                        <span>Settings</span>
+                    </a>
+                    <a href="#logs" class="menu-item" data-page="logs">
+                        <i class="fas fa-file-alt"></i>
+                        <span>System Logs</span>
+                    </a>
+                </div>
+            </div>
+            
+            <div class="sidebar-footer">
+                <div class="user-info">
+                    <div class="user-avatar">
+                        <i class="fas fa-user-circle"></i>
+                    </div>
+                    <div class="user-details">
+                        <span class="user-name">Administrator</span>
+                        <span class="user-role">Super Admin</span>
+                    </div>
+                </div>
+                <button class="logout-btn" id="logout-btn">
+                    <i class="fas fa-sign-out-alt"></i>
+                </button>
+            </div>
+        </nav>
+
+        <!-- Main Content -->
+        <main class="main-content">
+            <!-- Top Navigation -->
+            <header class="top-nav">
+                <div class="nav-left">
+                    <h1 class="page-title" id="page-title">Dashboard Overview</h1>
+                </div>
+                <div class="nav-right">
+                    <div class="nav-item">
+                        <i class="fas fa-bell"></i>
+                        <span class="badge">3</span>
+                    </div>
+                    <div class="nav-item">
+                        <i class="fas fa-sync-alt" id="refresh-btn"></i>
+                    </div>
+                    <div class="nav-item">
+                        <span id="current-time">Loading...</span>
+                    </div>
+                </div>
+            </header>
+
+            <!-- Page Content -->
+            <div class="page-content" id="page-content">
+                <!-- Overview page will be loaded here -->
+            </div>
+        </main>
+    </div>
+
+    <!-- WebSocket Status Indicator -->
+    <div class="ws-status" id="ws-status">
+        <span class="ws-dot"></span>
+        <span class="ws-text">Connecting...</span>
+    </div>
+
+    <!-- Scripts -->
+    <script src="/js/auth.js"></script>
+    <script src="/js/dashboard.js"></script>
+    <script src="/js/websocket.js"></script>
+    <script src="/js/charts.js"></script>
+    <script src="/js/pages/overview.js"></script>
+    <script src="/js/pages/analytics.js"></script>
+    <script src="/js/pages/costs.js"></script>
+    <script src="/js/pages/clients.js"></script>
+    <script src="/js/pages/providers.js"></script>
+    <script src="/js/pages/monitoring.js"></script>
+    <script src="/js/pages/settings.js"></script>
+    <script src="/js/pages/logs.js"></script>
+</body>
+</html>

static/js/auth.js

@@ -0,0 +1,269 @@
+// Authentication Module for LLM Proxy Dashboard
+
+class AuthManager {
+    constructor() {
+        this.isAuthenticated = false;
+        this.token = null;
+        this.user = null;
+        this.init();
+    }
+
+    init() {
+        // Check for existing session
+        const savedToken = localStorage.getItem('dashboard_token');
+        const savedUser = localStorage.getItem('dashboard_user');
+        
+        if (savedToken && savedUser) {
+            this.token = savedToken;
+            this.user = JSON.parse(savedUser);
+            this.isAuthenticated = true;
+            this.showDashboard();
+        } else {
+            this.showLogin();
+        }
+
+        // Setup login form
+        this.setupLoginForm();
+        this.setupLogout();
+    }
+
+    setupLoginForm() {
+        const loginForm = document.getElementById('login-form');
+        if (!loginForm) return;
+
+        loginForm.addEventListener('submit', async (e) => {
+            e.preventDefault();
+            
+            const username = document.getElementById('username').value;
+            const password = document.getElementById('password').value;
+            
+            await this.login(username, password);
+        });
+    }
+
+    setupLogout() {
+        const logoutBtn = document.getElementById('logout-btn');
+        if (!logoutBtn) return;
+
+        logoutBtn.addEventListener('click', () => {
+            this.logout();
+        });
+    }
+
+    async login(username, password) {
+        const errorElement = document.getElementById('login-error');
+        const loginBtn = document.querySelector('.login-btn');
+        
+        try {
+            // Show loading state
+            loginBtn.innerHTML = '<i class="fas fa-spinner fa-spin"></i> Authenticating...';
+            loginBtn.disabled = true;
+            
+            // Simple authentication - in production, this would call an API
+            // For now, using mock authentication
+            await new Promise(resolve => setTimeout(resolve, 1000));
+            
+            if (username === 'admin' && password === 'admin123') {
+                // Successful login
+                this.token = this.generateToken();
+                this.user = {
+                    username: 'admin',
+                    name: 'Administrator',
+                    role: 'Super Admin',
+                    avatar: null
+                };
+                
+                // Save to localStorage
+                localStorage.setItem('dashboard_token', this.token);
+                localStorage.setItem('dashboard_user', JSON.stringify(this.user));
+                
+                this.isAuthenticated = true;
+                this.showDashboard();
+                
+                // Show success message
+                this.showToast('Successfully logged in!', 'success');
+            } else {
+                throw new Error('Invalid credentials');
+            }
+        } catch (error) {
+            // Show error
+            errorElement.style.display = 'flex';
+            errorElement.querySelector('span').textContent = error.message;
+            
+            // Reset button
+            loginBtn.innerHTML = '<i class="fas fa-sign-in-alt"></i> Sign In';
+            loginBtn.disabled = false;
+        }
+    }
+
+    logout() {
+        // Clear localStorage
+        localStorage.removeItem('dashboard_token');
+        localStorage.removeItem('dashboard_user');
+        
+        // Reset state
+        this.isAuthenticated = false;
+        this.token = null;
+        this.user = null;
+        
+        // Show login screen
+        this.showLogin();
+        
+        // Show logout message
+        this.showToast('Successfully logged out', 'info');
+    }
+
+    generateToken() {
+        // Generate a simple token for demo purposes
+        // In production, this would come from the server
+        const timestamp = Date.now();
+        const random = Math.random().toString(36).substring(2);
+        return btoa(`${timestamp}:${random}`).replace(/=/g, '');
+    }
+
+    showLogin() {
+        const loginScreen = document.getElementById('login-screen');
+        const dashboard = document.getElementById('dashboard');
+        
+        if (loginScreen) loginScreen.style.display = 'flex';
+        if (dashboard) dashboard.style.display = 'none';
+        
+        // Clear form
+        const loginForm = document.getElementById('login-form');
+        if (loginForm) loginForm.reset();
+        
+        // Hide error
+        const errorElement = document.getElementById('login-error');
+        if (errorElement) errorElement.style.display = 'none';
+        
+        // Reset button
+        const loginBtn = document.querySelector('.login-btn');
+        if (loginBtn) {
+            loginBtn.innerHTML = '<i class="fas fa-sign-in-alt"></i> Sign In';
+            loginBtn.disabled = false;
+        }
+    }
+
+    showDashboard() {
+        const loginScreen = document.getElementById('login-screen');
+        const dashboard = document.getElementById('dashboard');
+        
+        if (loginScreen) loginScreen.style.display = 'none';
+        if (dashboard) dashboard.style.display = 'flex';
+        
+        // Update user info in sidebar
+        this.updateUserInfo();
+        
+        // Initialize dashboard components
+        if (typeof window.initDashboard === 'function') {
+            window.initDashboard();
+        }
+    }
+
+    updateUserInfo() {
+        const userNameElement = document.querySelector('.user-name');
+        const userRoleElement = document.querySelector('.user-role');
+        
+        if (userNameElement && this.user) {
+            userNameElement.textContent = this.user.name;
+        }
+        
+        if (userRoleElement && this.user) {
+            userRoleElement.textContent = this.user.role;
+        }
+    }
+
+    getAuthHeaders() {
+        if (!this.token) return {};
+        
+        return {
+            'Authorization': `Bearer ${this.token}`,
+            'Content-Type': 'application/json'
+        };
+    }
+
+    async fetchWithAuth(url, options = {}) {
+        const headers = this.getAuthHeaders();
+        
+        const response = await fetch(url, {
+            ...options,
+            headers: {
+                ...headers,
+                ...options.headers
+            }
+        });
+        
+        if (response.status === 401) {
+            // Token expired or invalid
+            this.logout();
+            throw new Error('Authentication required');
+        }
+        
+        return response;
+    }
+
+    showToast(message, type = 'info') {
+        // Create toast container if it doesn't exist
+        let container = document.querySelector('.toast-container');
+        if (!container) {
+            container = document.createElement('div');
+            container.className = 'toast-container';
+            document.body.appendChild(container);
+        }
+        
+        // Create toast
+        const toast = document.createElement('div');
+        toast.className = `toast ${type}`;
+        
+        // Set icon based on type
+        let icon = 'info-circle';
+        switch (type) {
+            case 'success':
+                icon = 'check-circle';
+                break;
+            case 'error':
+                icon = 'exclamation-circle';
+                break;
+            case 'warning':
+                icon = 'exclamation-triangle';
+                break;
+        }
+        
+        toast.innerHTML = `
+            <i class="fas fa-${icon} toast-icon"></i>
+            <div class="toast-content">
+                <div class="toast-title">${type.charAt(0).toUpperCase() + type.slice(1)}</div>
+                <div class="toast-message">${message}</div>
+            </div>
+            <button class="toast-close">
+                <i class="fas fa-times"></i>
+            </button>
+        `;
+        
+        // Add close functionality
+        const closeBtn = toast.querySelector('.toast-close');
+        closeBtn.addEventListener('click', () => {
+            toast.remove();
+        });
+        
+        // Add to container
+        container.appendChild(toast);
+        
+        // Auto-remove after 5 seconds
+        setTimeout(() => {
+            if (toast.parentNode) {
+                toast.remove();
+            }
+        }, 5000);
+    }
+}
+
+// Initialize auth manager when DOM is loaded
+document.addEventListener('DOMContentLoaded', () => {
+    window.authManager = new AuthManager();
+});
+
+// Export for use in other modules
+if (typeof module !== 'undefined' && module.exports) {
+    module.exports = AuthManager;
+}

static/js/charts.js

@@ -0,0 +1,533 @@
+// Chart.js Configuration and Helpers
+
+class ChartManager {
+    constructor() {
+        this.charts = new Map();
+        this.defaultColors = [
+            '#3b82f6', // Blue
+            '#10b981', // Green
+            '#f59e0b', // Yellow
+            '#ef4444', // Red
+            '#8b5cf6', // Purple
+            '#ec4899', // Pink
+            '#06b6d4', // Cyan
+            '#84cc16', // Lime
+            '#f97316', // Orange
+            '#6366f1', // Indigo
+        ];
+        this.init();
+    }
+
+    init() {
+        // Register Chart.js plugins if needed
+        this.registerPlugins();
+    }
+
+    registerPlugins() {
+        // Register a plugin for tooltip background
+        Chart.register({
+            id: 'customTooltip',
+            beforeDraw: (chart) => {
+                if (chart.tooltip._active && chart.tooltip._active.length) {
+                    const ctx = chart.ctx;
+                    const activePoint = chart.tooltip._active[0];
+                    const x = activePoint.element.x;
+                    const topY = chart.scales.y.top;
+                    const bottomY = chart.scales.y.bottom;
+
+                    ctx.save();
+                    ctx.beginPath();
+                    ctx.setLineDash([5, 5]);
+                    ctx.moveTo(x, topY);
+                    ctx.lineTo(x, bottomY);
+                    ctx.lineWidth = 1;
+                    ctx.strokeStyle = 'rgba(0, 0, 0, 0.1)';
+                    ctx.stroke();
+                    ctx.restore();
+                }
+            }
+        });
+    }
+
+    createChart(canvasId, config) {
+        const canvas = document.getElementById(canvasId);
+        if (!canvas) {
+            console.warn(`Canvas element #${canvasId} not found`);
+            return null;
+        }
+
+        // Destroy existing chart if it exists
+        if (this.charts.has(canvasId)) {
+            this.charts.get(canvasId).destroy();
+        }
+
+        // Create new chart
+        const ctx = canvas.getContext('2d');
+        const chart = new Chart(ctx, {
+            ...config,
+            options: {
+                responsive: true,
+                maintainAspectRatio: false,
+                plugins: {
+                    legend: {
+                        position: 'top',
+                        labels: {
+                            padding: 20,
+                            usePointStyle: true,
+                            pointStyle: 'circle'
+                        }
+                    },
+                    tooltip: {
+                        mode: 'index',
+                        intersect: false,
+                        backgroundColor: 'rgba(255, 255, 255, 0.95)',
+                        titleColor: '#1e293b',
+                        bodyColor: '#1e293b',
+                        borderColor: '#e2e8f0',
+                        borderWidth: 1,
+                        cornerRadius: 6,
+                        padding: 12,
+                        boxPadding: 6,
+                        callbacks: {
+                            label: function(context) {
+                                let label = context.dataset.label || '';
+                                if (label) {
+                                    label += ': ';
+                                }
+                                if (context.parsed.y !== null) {
+                                    label += context.parsed.y.toLocaleString();
+                                }
+                                return label;
+                            }
+                        }
+                    }
+                },
+                interaction: {
+                    intersect: false,
+                    mode: 'nearest'
+                },
+                scales: {
+                    x: {
+                        grid: {
+                            display: true,
+                            color: 'rgba(0, 0, 0, 0.05)'
+                        },
+                        ticks: {
+                            color: '#64748b'
+                        }
+                    },
+                    y: {
+                        beginAtZero: true,
+                        grid: {
+                            display: true,
+                            color: 'rgba(0, 0, 0, 0.05)'
+                        },
+                        ticks: {
+                            color: '#64748b',
+                            callback: function(value) {
+                                return value.toLocaleString();
+                            }
+                        }
+                    }
+                },
+                ...config.options
+            }
+        });
+
+        // Store chart reference
+        this.charts.set(canvasId, chart);
+
+        return chart;
+    }
+
+    destroyChart(canvasId) {
+        if (this.charts.has(canvasId)) {
+            this.charts.get(canvasId).destroy();
+            this.charts.delete(canvasId);
+        }
+    }
+
+    destroyAllCharts() {
+        this.charts.forEach((chart, canvasId) => {
+            chart.destroy();
+        });
+        this.charts.clear();
+    }
+
+    // Chart templates
+    createLineChart(canvasId, data, options = {}) {
+        const config = {
+            type: 'line',
+            data: {
+                labels: data.labels || [],
+                datasets: data.datasets.map((dataset, index) => ({
+                    label: dataset.label,
+                    data: dataset.data,
+                    borderColor: dataset.color || this.defaultColors[index % this.defaultColors.length],
+                    backgroundColor: dataset.fill ? this.hexToRgba(dataset.color || this.defaultColors[index % this.defaultColors.length], 0.1) : 'transparent',
+                    borderWidth: 2,
+                    pointRadius: 3,
+                    pointHoverRadius: 6,
+                    pointBackgroundColor: dataset.color || this.defaultColors[index % this.defaultColors.length],
+                    pointBorderColor: '#ffffff',
+                    pointBorderWidth: 2,
+                    fill: dataset.fill || false,
+                    tension: 0.4,
+                    ...dataset
+                }))
+            },
+            options: {
+                plugins: {
+                    tooltip: {
+                        callbacks: {
+                            label: function(context) {
+                                let label = context.dataset.label || '';
+                                if (label) {
+                                    label += ': ';
+                                }
+                                if (context.parsed.y !== null) {
+                                    label += context.parsed.y.toLocaleString();
+                                }
+                                return label;
+                            }
+                        }
+                    }
+                },
+                ...options
+            }
+        };
+
+        return this.createChart(canvasId, config);
+    }
+
+    createBarChart(canvasId, data, options = {}) {
+        const config = {
+            type: 'bar',
+            data: {
+                labels: data.labels || [],
+                datasets: data.datasets.map((dataset, index) => ({
+                    label: dataset.label,
+                    data: dataset.data,
+                    backgroundColor: dataset.color || this.defaultColors[index % this.defaultColors.length],
+                    borderColor: dataset.borderColor || '#ffffff',
+                    borderWidth: 1,
+                    borderRadius: 4,
+                    borderSkipped: false,
+                    ...dataset
+                }))
+            },
+            options: {
+                plugins: {
+                    tooltip: {
+                        callbacks: {
+                            label: function(context) {
+                                let label = context.dataset.label || '';
+                                if (label) {
+                                    label += ': ';
+                                }
+                                if (context.parsed.y !== null) {
+                                    label += context.parsed.y.toLocaleString();
+                                }
+                                return label;
+                            }
+                        }
+                    }
+                },
+                ...options
+            }
+        };
+
+        return this.createChart(canvasId, config);
+    }
+
+    createPieChart(canvasId, data, options = {}) {
+        const config = {
+            type: 'pie',
+            data: {
+                labels: data.labels || [],
+                datasets: [{
+                    data: data.data || [],
+                    backgroundColor: data.colors || this.defaultColors.slice(0, data.data.length),
+                    borderColor: '#ffffff',
+                    borderWidth: 2,
+                    hoverOffset: 15
+                }]
+            },
+            options: {
+                plugins: {
+                    legend: {
+                        position: 'right',
+                        labels: {
+                            padding: 20,
+                            usePointStyle: true,
+                            pointStyle: 'circle'
+                        }
+                    },
+                    tooltip: {
+                        callbacks: {
+                            label: function(context) {
+                                const label = context.label || '';
+                                const value = context.raw || 0;
+                                const total = context.dataset.data.reduce((a, b) => a + b, 0);
+                                const percentage = Math.round((value / total) * 100);
+                                return `${label}: ${value.toLocaleString()} (${percentage}%)`;
+                            }
+                        }
+                    }
+                },
+                ...options
+            }
+        };
+
+        return this.createChart(canvasId, config);
+    }
+
+    createDoughnutChart(canvasId, data, options = {}) {
+        const config = {
+            type: 'doughnut',
+            data: {
+                labels: data.labels || [],
+                datasets: [{
+                    data: data.data || [],
+                    backgroundColor: data.colors || this.defaultColors.slice(0, data.data.length),
+                    borderColor: '#ffffff',
+                    borderWidth: 2,
+                    hoverOffset: 15
+                }]
+            },
+            options: {
+                cutout: '60%',
+                plugins: {
+                    legend: {
+                        position: 'right',
+                        labels: {
+                            padding: 20,
+                            usePointStyle: true,
+                            pointStyle: 'circle'
+                        }
+                    },
+                    tooltip: {
+                        callbacks: {
+                            label: function(context) {
+                                const label = context.label || '';
+                                const value = context.raw || 0;
+                                const total = context.dataset.data.reduce((a, b) => a + b, 0);
+                                const percentage = Math.round((value / total) * 100);
+                                return `${label}: ${value.toLocaleString()} (${percentage}%)`;
+                            }
+                        }
+                    }
+                },
+                ...options
+            }
+        };
+
+        return this.createChart(canvasId, config);
+    }
+
+    createHorizontalBarChart(canvasId, data, options = {}) {
+        const config = {
+            type: 'bar',
+            data: {
+                labels: data.labels || [],
+                datasets: data.datasets.map((dataset, index) => ({
+                    label: dataset.label,
+                    data: dataset.data,
+                    backgroundColor: dataset.color || this.defaultColors[index % this.defaultColors.length],
+                    borderColor: dataset.borderColor || '#ffffff',
+                    borderWidth: 1,
+                    borderRadius: 4,
+                    ...dataset
+                }))
+            },
+            options: {
+                indexAxis: 'y',
+                plugins: {
+                    legend: {
+                        display: false
+                    },
+                    tooltip: {
+                        callbacks: {
+                            label: function(context) {
+                                let label = context.dataset.label || '';
+                                if (label) {
+                                    label += ': ';
+                                }
+                                if (context.parsed.x !== null) {
+                                    label += context.parsed.x.toLocaleString();
+                                }
+                                return label;
+                            }
+                        }
+                    }
+                },
+                scales: {
+                    x: {
+                        beginAtZero: true,
+                        grid: {
+                            display: true,
+                            color: 'rgba(0, 0, 0, 0.05)'
+                        },
+                        ticks: {
+                            color: '#64748b',
+                            callback: function(value) {
+                                return value.toLocaleString();
+                            }
+                        }
+                    },
+                    y: {
+                        grid: {
+                            display: false
+                        },
+                        ticks: {
+                            color: '#64748b'
+                        }
+                    }
+                },
+                ...options
+            }
+        };
+
+        return this.createChart(canvasId, config);
+    }
+
+    // Helper methods
+    hexToRgba(hex, alpha = 1) {
+        const r = parseInt(hex.slice(1, 3), 16);
+        const g = parseInt(hex.slice(3, 5), 16);
+        const b = parseInt(hex.slice(5, 7), 16);
+        return `rgba(${r}, ${g}, ${b}, ${alpha})`;
+    }
+
+    generateTimeLabels(hours = 24) {
+        const now = luxon.DateTime.now();
+        const labels = [];
+        
+        for (let i = hours - 1; i >= 0; i--) {
+            const time = now.minus({ hours: i });
+            labels.push(time.toFormat('HH:00'));
+        }
+        
+        return labels;
+    }
+
+    generateDateLabels(days = 7) {
+        const now = luxon.DateTime.now();
+        const labels = [];
+        
+        for (let i = days - 1; i >= 0; i--) {
+            const date = now.minus({ days: i });
+            labels.push(date.toFormat('MMM dd'));
+        }
+        
+        return labels;
+    }
+
+    // Demo data generators
+    generateDemoTimeSeries(hours = 24, seriesCount = 1) {
+        const labels = this.generateTimeLabels(hours);
+        const datasets = [];
+        
+        for (let i = 0; i < seriesCount; i++) {
+            const data = [];
+            let value = Math.floor(Math.random() * 100) + 50;
+            
+            for (let j = 0; j < hours; j++) {
+                // Add some randomness but keep trend
+                value += Math.floor(Math.random() * 20) - 10;
+                value = Math.max(10, value);
+                data.push(value);
+            }
+            
+            datasets.push({
+                label: `Series ${i + 1}`,
+                data: data,
+                color: this.defaultColors[i % this.defaultColors.length]
+            });
+        }
+        
+        return { labels, datasets };
+    }
+
+    generateDemoBarData(labels, seriesCount = 1) {
+        const datasets = [];
+        
+        for (let i = 0; i < seriesCount; i++) {
+            const data = labels.map(() => Math.floor(Math.random() * 100) + 20);
+            
+            datasets.push({
+                label: `Dataset ${i + 1}`,
+                data: data,
+                color: this.defaultColors[i % this.defaultColors.length]
+            });
+        }
+        
+        return { labels, datasets };
+    }
+
+    generateDemoPieData(labels) {
+        const data = labels.map(() => Math.floor(Math.random() * 100) + 10);
+        const total = data.reduce((a, b) => a + b, 0);
+        
+        return {
+            labels: labels,
+            data: data,
+            colors: labels.map((_, i) => this.defaultColors[i % this.defaultColors.length])
+        };
+    }
+
+    // Update chart data
+    updateChartData(canvasId, newData) {
+        const chart = this.charts.get(canvasId);
+        if (!chart) return;
+        
+        chart.data.labels = newData.labels || chart.data.labels;
+        
+        if (newData.datasets) {
+            newData.datasets.forEach((dataset, index) => {
+                if (chart.data.datasets[index]) {
+                    chart.data.datasets[index].data = dataset.data;
+                    if (dataset.label) {
+                        chart.data.datasets[index].label = dataset.label;
+                    }
+                }
+            });
+        }
+        
+        chart.update();
+    }
+
+    // Add data point to time series
+    addDataPoint(canvasId, newPoint, datasetIndex = 0) {
+        const chart = this.charts.get(canvasId);
+        if (!chart || chart.config.type !== 'line') return;
+        
+        const dataset = chart.data.datasets[datasetIndex];
+        if (!dataset) return;
+        
+        // Add new point
+        dataset.data.push(newPoint);
+        
+        // Remove oldest point if we have too many
+        if (dataset.data.length > 100) {
+            dataset.data.shift();
+            chart.data.labels.shift();
+        } else {
+            // Add new label
+            const now = luxon.DateTime.now();
+            chart.data.labels.push(now.toFormat('HH:mm:ss'));
+        }
+        
+        chart.update();
+    }
+}
+
+// Initialize chart manager when DOM is loaded
+document.addEventListener('DOMContentLoaded', () => {
+    window.chartManager = new ChartManager();
+});
+
+// Export for use in other modules
+if (typeof module !== 'undefined' && module.exports) {
+    module.exports = ChartManager;
+}

static/js/dashboard.js

@@ -0,0 +1,872 @@
+// Main Dashboard Controller
+
+class Dashboard {
+    constructor() {
+        this.currentPage = 'overview';
+        this.pages = {};
+        this.init();
+    }
+
+    init() {
+        // Initialize only if authenticated
+        if (!window.authManager || !window.authManager.isAuthenticated) {
+            return;
+        }
+
+        this.setupNavigation();
+        this.setupSidebar();
+        this.setupRefresh();
+        this.updateTime();
+        this.loadPage(this.currentPage);
+        
+        // Start time updates
+        setInterval(() => this.updateTime(), 1000);
+    }
+
+    setupNavigation() {
+        // Handle menu item clicks
+        const menuItems = document.querySelectorAll('.menu-item');
+        menuItems.forEach(item => {
+            item.addEventListener('click', (e) => {
+                e.preventDefault();
+                
+                // Get page from data attribute or href
+                const page = item.getAttribute('data-page') || 
+                            item.getAttribute('href').substring(1);
+                
+                // Update active state
+                menuItems.forEach(i => i.classList.remove('active'));
+                item.classList.add('active');
+                
+                // Load page
+                this.loadPage(page);
+            });
+        });
+
+        // Handle hash changes (browser back/forward)
+        window.addEventListener('hashchange', () => {
+            const page = window.location.hash.substring(1) || 'overview';
+            this.loadPage(page);
+        });
+    }
+
+    setupSidebar() {
+        const toggleBtn = document.getElementById('sidebar-toggle');
+        const sidebar = document.querySelector('.sidebar');
+        
+        if (toggleBtn && sidebar) {
+            toggleBtn.addEventListener('click', () => {
+                sidebar.classList.toggle('collapsed');
+                
+                // Save preference
+                const isCollapsed = sidebar.classList.contains('collapsed');
+                localStorage.setItem('sidebar_collapsed', isCollapsed);
+            });
+            
+            // Load saved preference
+            const savedState = localStorage.getItem('sidebar_collapsed');
+            if (savedState === 'true') {
+                sidebar.classList.add('collapsed');
+            }
+        }
+    }
+
+    setupRefresh() {
+        const refreshBtn = document.getElementById('refresh-btn');
+        if (refreshBtn) {
+            refreshBtn.addEventListener('click', () => {
+                this.refreshCurrentPage();
+            });
+        }
+    }
+
+    updateTime() {
+        const timeElement = document.getElementById('current-time');
+        if (!timeElement) return;
+        
+        const now = luxon.DateTime.now();
+        timeElement.textContent = now.toFormat('HH:mm:ss');
+    }
+
+    async loadPage(page) {
+        // Update current page
+        this.currentPage = page;
+        
+        // Update URL hash
+        window.location.hash = page;
+        
+        // Update page title
+        this.updatePageTitle(page);
+        
+        // Show loading state
+        this.showLoading();
+        
+        try {
+            // Load page content
+            await this.loadPageContent(page);
+            
+            // Initialize page-specific functionality
+            await this.initializePage(page);
+            
+        } catch (error) {
+            console.error(`Error loading page ${page}:`, error);
+            this.showError(`Failed to load ${page} page`);
+        } finally {
+            // Hide loading state
+            this.hideLoading();
+        }
+    }
+
+    updatePageTitle(page) {
+        const titleElement = document.getElementById('page-title');
+        if (!titleElement) return;
+        
+        const titles = {
+            'overview': 'Dashboard Overview',
+            'analytics': 'Usage Analytics',
+            'costs': 'Cost Management',
+            'clients': 'Client Management',
+            'providers': 'Provider Configuration',
+            'monitoring': 'Real-time Monitoring',
+            'settings': 'System Settings',
+            'logs': 'System Logs'
+        };
+        
+        titleElement.textContent = titles[page] || 'Dashboard';
+    }
+
+    showLoading() {
+        const content = document.getElementById('page-content');
+        if (!content) return;
+        
+        content.classList.add('loading');
+    }
+
+    hideLoading() {
+        const content = document.getElementById('page-content');
+        if (!content) return;
+        
+        content.classList.remove('loading');
+    }
+
+    async loadPageContent(page) {
+        const content = document.getElementById('page-content');
+        if (!content) return;
+        
+        // For now, we'll generate content dynamically
+        // In a real app, you might fetch HTML templates or use a framework
+        
+        let html = '';
+        
+        switch (page) {
+            case 'overview':
+                html = await this.getOverviewContent();
+                break;
+            case 'analytics':
+                html = await this.getAnalyticsContent();
+                break;
+            case 'costs':
+                html = await this.getCostsContent();
+                break;
+            case 'clients':
+                html = await this.getClientsContent();
+                break;
+            case 'providers':
+                html = await this.getProvidersContent();
+                break;
+            case 'monitoring':
+                html = await this.getMonitoringContent();
+                break;
+            case 'settings':
+                html = await this.getSettingsContent();
+                break;
+            case 'logs':
+                html = await this.getLogsContent();
+                break;
+            default:
+                html = '<div class="empty-state"><h3>Page not found</h3></div>';
+        }
+        
+        content.innerHTML = html;
+    }
+
+    async initializePage(page) {
+        // Initialize page-specific JavaScript
+        switch (page) {
+            case 'overview':
+                if (typeof window.initOverview === 'function') {
+                    await window.initOverview();
+                }
+                break;
+            case 'analytics':
+                if (typeof window.initAnalytics === 'function') {
+                    await window.initAnalytics();
+                }
+                break;
+            case 'costs':
+                if (typeof window.initCosts === 'function') {
+                    await window.initCosts();
+                }
+                break;
+            case 'clients':
+                if (typeof window.initClients === 'function') {
+                    await window.initClients();
+                }
+                break;
+            case 'providers':
+                if (typeof window.initProviders === 'function') {
+                    await window.initProviders();
+                }
+                break;
+            case 'monitoring':
+                if (typeof window.initMonitoring === 'function') {
+                    await window.initMonitoring();
+                }
+                break;
+            case 'settings':
+                if (typeof window.initSettings === 'function') {
+                    await window.initSettings();
+                }
+                break;
+            case 'logs':
+                if (typeof window.initLogs === 'function') {
+                    await window.initLogs();
+                }
+                break;
+        }
+    }
+
+    refreshCurrentPage() {
+        this.loadPage(this.currentPage);
+        
+        // Show refresh animation
+        const refreshBtn = document.getElementById('refresh-btn');
+        if (refreshBtn) {
+            refreshBtn.classList.add('fa-spin');
+            setTimeout(() => {
+                refreshBtn.classList.remove('fa-spin');
+            }, 1000);
+        }
+        
+        // Show toast notification
+        if (window.authManager) {
+            window.authManager.showToast('Page refreshed', 'success');
+        }
+    }
+
+    showError(message) {
+        const content = document.getElementById('page-content');
+        if (!content) return;
+        
+        content.innerHTML = `
+            <div class="empty-state">
+                <i class="fas fa-exclamation-triangle"></i>
+                <h3>Error</h3>
+                <p>${message}</p>
+                <button class="btn btn-primary" onclick="window.dashboard.refreshCurrentPage()">
+                    <i class="fas fa-redo"></i> Try Again
+                </button>
+            </div>
+        `;
+    }
+
+    // Page content generators
+    async getOverviewContent() {
+        return `
+            <div class="stats-grid" id="overview-stats">
+                <!-- Stats will be loaded dynamically -->
+            </div>
+            
+            <div class="chart-container">
+                <div class="chart-header">
+                    <h3 class="chart-title">Request Volume (Last 24 Hours)</h3>
+                    <div class="chart-controls">
+                        <button class="chart-control-btn active" data-period="24h">24H</button>
+                        <button class="chart-control-btn" data-period="7d">7D</button>
+                        <button class="chart-control-btn" data-period="30d">30D</button>
+                    </div>
+                </div>
+                <canvas id="requests-chart" height="300"></canvas>
+            </div>
+            
+            <div class="grid-2">
+                <div class="chart-container">
+                    <div class="chart-header">
+                        <h3 class="chart-title">Provider Distribution</h3>
+                    </div>
+                    <canvas id="providers-chart" height="250"></canvas>
+                </div>
+                
+                <div class="chart-container">
+                    <div class="chart-header">
+                        <h3 class="chart-title">System Health</h3>
+                    </div>
+                    <div id="system-health">
+                        <!-- Health indicators will be loaded dynamically -->
+                    </div>
+                </div>
+            </div>
+            
+            <div class="card">
+                <div class="card-header">
+                    <div>
+                        <h3 class="card-title">Recent Requests</h3>
+                        <p class="card-subtitle">Last 50 requests</p>
+                    </div>
+                    <div class="card-actions">
+                        <button class="card-action-btn" title="Refresh">
+                            <i class="fas fa-redo"></i>
+                        </button>
+                    </div>
+                </div>
+                <div class="table-container">
+                    <table class="table" id="recent-requests">
+                        <thead>
+                            <tr>
+                                <th>Time</th>
+                                <th>Client</th>
+                                <th>Provider</th>
+                                <th>Model</th>
+                                <th>Tokens</th>
+                                <th>Status</th>
+                            </tr>
+                        </thead>
+                        <tbody>
+                            <!-- Requests will be loaded dynamically -->
+                        </tbody>
+                    </table>
+                </div>
+            </div>
+        `;
+    }
+
+    async getAnalyticsContent() {
+        return `
+            <div class="card">
+                <div class="card-header">
+                    <div>
+                        <h3 class="card-title">Usage Analytics</h3>
+                        <p class="card-subtitle">Filter and analyze usage data</p>
+                    </div>
+                    <div class="card-actions">
+                        <button class="btn btn-secondary">
+                            <i class="fas fa-download"></i> Export
+                        </button>
+                    </div>
+                </div>
+                <div class="card-body">
+                    <div class="form-row">
+                        <div class="form-control">
+                            <label>Date Range</label>
+                            <select id="date-range">
+                                <option value="24h">Last 24 Hours</option>
+                                <option value="7d" selected>Last 7 Days</option>
+                                <option value="30d">Last 30 Days</option>
+                                <option value="custom">Custom Range</option>
+                            </select>
+                        </div>
+                        <div class="form-control">
+                            <label>Client</label>
+                            <select id="client-filter">
+                                <option value="all">All Clients</option>
+                                <!-- Client options will be loaded dynamically -->
+                            </select>
+                        </div>
+                        <div class="form-control">
+                            <label>Provider</label>
+                            <select id="provider-filter">
+                                <option value="all">All Providers</option>
+                                <option value="openai">OpenAI</option>
+                                <option value="gemini">Gemini</option>
+                                <option value="deepseek">DeepSeek</option>
+                                <option value="grok">Grok</option>
+                            </select>
+                        </div>
+                    </div>
+                </div>
+            </div>
+            
+            <div class="chart-container">
+                <div class="chart-header">
+                    <h3 class="chart-title">Request Trends</h3>
+                    <div class="chart-controls">
+                        <button class="chart-control-btn active" data-metric="requests">Requests</button>
+                        <button class="chart-control-btn" data-metric="tokens">Tokens</button>
+                        <button class="chart-control-btn" data-metric="cost">Cost</button>
+                    </div>
+                </div>
+                <canvas id="analytics-chart" height="350"></canvas>
+            </div>
+            
+            <div class="grid-2">
+                <div class="chart-container">
+                    <div class="chart-header">
+                        <h3 class="chart-title">Top Clients</h3>
+                    </div>
+                    <canvas id="clients-chart" height="300"></canvas>
+                </div>
+                
+                <div class="chart-container">
+                    <div class="chart-header">
+                        <h3 class="chart-title">Top Models</h3>
+                    </div>
+                    <canvas id="models-chart" height="300"></canvas>
+                </div>
+            </div>
+            
+            <div class="card">
+                <div class="card-header">
+                    <h3 class="card-title">Detailed Usage Data</h3>
+                </div>
+                <div class="table-container">
+                    <table class="table" id="usage-table">
+                        <thead>
+                            <tr>
+                                <th>Date</th>
+                                <th>Client</th>
+                                <th>Provider</th>
+                                <th>Model</th>
+                                <th>Requests</th>
+                                <th>Tokens</th>
+                                <th>Cost</th>
+                            </tr>
+                        </thead>
+                        <tbody>
+                            <!-- Usage data will be loaded dynamically -->
+                        </tbody>
+                    </table>
+                </div>
+            </div>
+        `;
+    }
+
+    async getCostsContent() {
+        return `
+            <div class="stats-grid" id="cost-stats">
+                <!-- Cost stats will be loaded dynamically -->
+            </div>
+            
+            <div class="chart-container">
+                <div class="chart-header">
+                    <h3 class="chart-title">Cost Breakdown</h3>
+                    <div class="chart-controls">
+                        <button class="chart-control-btn active" data-breakdown="provider">By Provider</button>
+                        <button class="chart-control-btn" data-breakdown="client">By Client</button>
+                        <button class="chart-control-btn" data-breakdown="model">By Model</button>
+                    </div>
+                </div>
+                <canvas id="costs-chart" height="300"></canvas>
+            </div>
+            
+            <div class="grid-2">
+                <div class="card">
+                    <div class="card-header">
+                        <h3 class="card-title">Budget Tracking</h3>
+                    </div>
+                    <div class="card-body">
+                        <div id="budget-progress">
+                            <!-- Budget progress bars will be loaded dynamically -->
+                        </div>
+                    </div>
+                </div>
+                
+                <div class="card">
+                    <div class="card-header">
+                        <h3 class="card-title">Cost Projections</h3>
+                    </div>
+                    <div class="card-body">
+                        <div id="cost-projections">
+                            <!-- Projections will be loaded dynamically -->
+                        </div>
+                    </div>
+                </div>
+            </div>
+            
+            <div class="card">
+                <div class="card-header">
+                    <div>
+                        <h3 class="card-title">Pricing Configuration</h3>
+                        <p class="card-subtitle">Current provider pricing</p>
+                    </div>
+                    <div class="card-actions">
+                        <button class="btn btn-primary" id="edit-pricing">
+                            <i class="fas fa-edit"></i> Edit
+                        </button>
+                    </div>
+                </div>
+                <div class="table-container">
+                    <table class="table" id="pricing-table">
+                        <thead>
+                            <tr>
+                                <th>Provider</th>
+                                <th>Model</th>
+                                <th>Input Price</th>
+                                <th>Output Price</th>
+                                <th>Last Updated</th>
+                            </tr>
+                        </thead>
+                        <tbody>
+                            <!-- Pricing data will be loaded dynamically -->
+                        </tbody>
+                    </table>
+                </div>
+            </div>
+        `;
+    }
+
+    async getClientsContent() {
+        return `
+            <div class="card">
+                <div class="card-header">
+                    <div>
+                        <h3 class="card-title">Client Management</h3>
+                        <p class="card-subtitle">Manage API clients and tokens</p>
+                    </div>
+                    <div class="card-actions">
+                        <button class="btn btn-primary" id="add-client">
+                            <i class="fas fa-plus"></i> Add Client
+                        </button>
+                    </div>
+                </div>
+                <div class="card-body">
+                    <div class="table-container">
+                        <table class="table" id="clients-table">
+                            <thead>
+                                <tr>
+                                    <th>Client ID</th>
+                                    <th>Name</th>
+                                    <th>Token</th>
+                                    <th>Created</th>
+                                    <th>Last Used</th>
+                                    <th>Requests</th>
+                                    <th>Status</th>
+                                    <th>Actions</th>
+                                </tr>
+                            </thead>
+                            <tbody>
+                                <!-- Clients will be loaded dynamically -->
+                            </tbody>
+                        </table>
+                    </div>
+                </div>
+            </div>
+            
+            <div class="grid-2">
+                <div class="card">
+                    <div class="card-header">
+                        <h3 class="card-title">Client Usage Summary</h3>
+                    </div>
+                    <div class="card-body">
+                        <canvas id="client-usage-chart" height="250"></canvas>
+                    </div>
+                </div>
+                
+                <div class="card">
+                    <div class="card-header">
+                        <h3 class="card-title">Rate Limit Status</h3>
+                    </div>
+                    <div class="card-body">
+                        <div id="rate-limit-status">
+                            <!-- Rate limit status will be loaded dynamically -->
+                        </div>
+                    </div>
+                </div>
+            </div>
+        `;
+    }
+
+    async getProvidersContent() {
+        return `
+            <div class="stats-grid" id="provider-stats">
+                <!-- Provider stats will be loaded dynamically -->
+            </div>
+            
+            <div class="card">
+                <div class="card-header">
+                    <h3 class="card-title">Provider Configuration</h3>
+                </div>
+                <div class="card-body">
+                    <div id="providers-list">
+                        <!-- Providers will be loaded dynamically -->
+                    </div>
+                </div>
+            </div>
+            
+            <div class="grid-2">
+                <div class="card">
+                    <div class="card-header">
+                        <h3 class="card-title">Model Availability</h3>
+                    </div>
+                    <div class="card-body">
+                        <div id="models-list">
+                            <!-- Models will be loaded dynamically -->
+                        </div>
+                    </div>
+                </div>
+                
+                <div class="card">
+                    <div class="card-header">
+                        <h3 class="card-title">Connection Tests</h3>
+                    </div>
+                    <div class="card-body">
+                        <div id="connection-tests">
+                            <!-- Test results will be loaded dynamically -->
+                        </div>
+                        <div class="form-actions">
+                            <button class="btn btn-primary" id="test-all-providers">
+                                <i class="fas fa-play"></i> Test All Providers
+                            </button>
+                        </div>
+                    </div>
+                </div>
+            </div>
+        `;
+    }
+
+    async getMonitoringContent() {
+        return `
+            <div class="card">
+                <div class="card-header">
+                    <div>
+                        <h3 class="card-title">Real-time Monitoring</h3>
+                        <p class="card-subtitle">Live request stream and system metrics</p>
+                    </div>
+                    <div class="card-actions">
+                        <button class="btn btn-secondary" id="pause-monitoring">
+                            <i class="fas fa-pause"></i> Pause
+                        </button>
+                    </div>
+                </div>
+                <div class="card-body">
+                    <div class="grid-2">
+                        <div>
+                            <h4>Live Request Stream</h4>
+                            <div id="request-stream" class="monitoring-stream">
+                                <!-- Live requests will appear here -->
+                            </div>
+                        </div>
+                        <div>
+                            <h4>System Metrics</h4>
+                            <div id="system-metrics">
+                                <!-- System metrics will be loaded dynamically -->
+                            </div>
+                        </div>
+                    </div>
+                </div>
+            </div>
+            
+            <div class="grid-3">
+                <div class="chart-container">
+                    <div class="chart-header">
+                        <h3 class="chart-title">Response Time (ms)</h3>
+                    </div>
+                    <canvas id="response-time-chart" height="200"></canvas>
+                </div>
+                
+                <div class="chart-container">
+                    <div class="chart-header">
+                        <h3 class="chart-title">Error Rate (%)</h3>
+                    </div>
+                    <canvas id="error-rate-chart" height="200"></canvas>
+                </div>
+                
+                <div class="chart-container">
+                    <div class="chart-header">
+                        <h3 class="chart-title">Rate Limit Usage</h3>
+                    </div>
+                    <canvas id="rate-limit-chart" height="200"></canvas>
+                </div>
+            </div>
+            
+            <div class="card">
+                <div class="card-header">
+                    <h3 class="card-title">System Logs (Live)</h3>
+                </div>
+                <div class="card-body">
+                    <div id="system-logs" class="log-stream">
+                        <!-- System logs will appear here -->
+                    </div>
+                </div>
+            </div>
+        `;
+    }
+
+    async getSettingsContent() {
+        return `
+            <div class="card">
+                <div class="card-header">
+                    <h3 class="card-title">System Settings</h3>
+                </div>
+                <div class="card-body">
+                    <form id="settings-form">
+                        <div class="form-section">
+                            <h4>General Configuration</h4>
+                            <div class="form-row">
+                                <div class="form-control">
+                                    <label>Server Port</label>
+                                    <input type="number" id="server-port" value="8080" min="1024" max="65535">
+                                </div>
+                                <div class="form-control">
+                                    <label>Log Level</label>
+                                    <select id="log-level">
+                                        <option value="error">Error</option>
+                                        <option value="warn">Warning</option>
+                                        <option value="info" selected>Info</option>
+                                        <option value="debug">Debug</option>
+                                    </select>
+                                </div>
+                            </div>
+                        </div>
+                        
+                        <div class="form-section">
+                            <h4>Database Settings</h4>
+                            <div class="form-row">
+                                <div class="form-control">
+                                    <label>Database Path</label>
+                                    <input type="text" id="db-path" value="./data/llm-proxy.db">
+                                </div>
+                                <div class="form-control">
+                                    <label>Backup Interval (hours)</label>
+                                    <input type="number" id="backup-interval" value="24" min="1" max="168">
+                                </div>
+                            </div>
+                        </div>
+                        
+                        <div class="form-section">
+                            <h4>Security Settings</h4>
+                            <div class="form-row">
+                                <div class="form-control">
+                                    <label>Dashboard Password</label>
+                                    <input type="password" id="dashboard-password" placeholder="Leave empty to keep current">
+                                </div>
+                                <div class="form-control">
+                                    <label>Session Timeout (minutes)</label>
+                                    <input type="number" id="session-timeout" value="30" min="5" max="1440">
+                                </div>
+                            </div>
+                        </div>
+                        
+                        <div class="form-actions">
+                            <button type="button" class="btn btn-secondary" id="reset-settings">
+                                <i class="fas fa-undo"></i> Reset
+                            </button>
+                            <button type="submit" class="btn btn-primary">
+                                <i class="fas fa-save"></i> Save Settings
+                            </button>
+                        </div>
+                    </form>
+                </div>
+            </div>
+            
+            <div class="grid-2">
+                <div class="card">
+                    <div class="card-header">
+                        <h3 class="card-title">Database Management</h3>
+                    </div>
+                    <div class="card-body">
+                        <div class="form-actions">
+                            <button class="btn btn-secondary" id="backup-db">
+                                <i class="fas fa-download"></i> Backup Database
+                            </button>
+                            <button class="btn btn-warning" id="optimize-db">
+                                <i class="fas fa-magic"></i> Optimize Database
+                            </button>
+                        </div>
+                    </div>
+                </div>
+                
+                <div class="card">
+                    <div class="card-header">
+                        <h3 class="card-title">System Information</h3>
+                    </div>
+                    <div class="card-body">
+                        <div id="system-info">
+                            <!-- System info will be loaded dynamically -->
+                        </div>
+                    </div>
+                </div>
+            </div>
+        `;
+    }
+
+    async getLogsContent() {
+        return `
+            <div class="card">
+                <div class="card-header">
+                    <div>
+                        <h3 class="card-title">System Logs</h3>
+                        <p class="card-subtitle">View and filter system logs</p>
+                    </div>
+                    <div class="card-actions">
+                        <button class="btn btn-secondary" id="download-logs">
+                            <i class="fas fa-download"></i> Download
+                        </button>
+                        <button class="btn btn-danger" id="clear-logs">
+                            <i class="fas fa-trash"></i> Clear
+                        </button>
+                    </div>
+                </div>
+                <div class="card-body">
+                    <div class="form-row">
+                        <div class="form-control">
+                            <label>Log Level</label>
+                            <select id="log-filter">
+                                <option value="all">All Levels</option>
+                                <option value="error">Error</option>
+                                <option value="warn">Warning</option>
+                                <option value="info">Info</option>
+                                <option value="debug">Debug</option>
+                            </select>
+                        </div>
+                        <div class="form-control">
+                            <label>Time Range</label>
+                            <select id="log-time-range">
+                                <option value="1h">Last Hour</option>
+                                <option value="24h" selected>Last 24 Hours</option>
+                                <option value="7d">Last 7 Days</option>
+                                <option value="30d">Last 30 Days</option>
+                            </select>
+                        </div>
+                        <div class="form-control">
+                            <label>Search</label>
+                            <input type="text" id="log-search" placeholder="Search logs...">
+                        </div>
+                    </div>
+                    
+                    <div class="table-container">
+                        <table class="table" id="logs-table">
+                            <thead>
+                                <tr>
+                                    <th>Timestamp</th>
+                                    <th>Level</th>
+                                    <th>Source</th>
+                                    <th>Message</th>
+                                </tr>
+                            </thead>
+                            <tbody>
+                                <!-- Logs will be loaded dynamically -->
+                            </tbody>
+                        </table>
+                    </div>
+                </div>
+            </div>
+        `;
+    }
+}
+
+// Initialize dashboard when DOM is loaded
+document.addEventListener('DOMContentLoaded', () => {
+    window.initDashboard = () => {
+        window.dashboard = new Dashboard();
+    };
+    
+    // If already authenticated, initialize immediately
+    if (window.authManager && window.authManager.isAuthenticated) {
+        window.initDashboard();
+    }
+});
+
+// Export for use in other modules
+if (typeof module !== 'undefined' && module.exports) {
+    module.exports = Dashboard;
+}

static/js/pages/analytics.js

@@ -0,0 +1,334 @@
+// Analytics Page Module
+
+class AnalyticsPage {
+    constructor() {
+        this.filters = {
+            dateRange: '7d',
+            client: 'all',
+            provider: 'all'
+        };
+        this.init();
+    }
+
+    async init() {
+        // Load initial data
+        await this.loadFilters();
+        await this.loadCharts();
+        await this.loadUsageData();
+        
+        // Setup event listeners
+        this.setupEventListeners();
+    }
+
+    async loadFilters() {
+        try {
+            // Load clients for filter dropdown
+            // In a real app, this would fetch from /api/clients
+            const clients = [
+                { id: 'client-1', name: 'Web Application' },
+                { id: 'client-2', name: 'Mobile App' },
+                { id: 'client-3', name: 'API Integration' },
+                { id: 'client-4', name: 'Internal Tools' },
+                { id: 'client-5', name: 'Testing Suite' }
+            ];
+            
+            this.renderClientFilter(clients);
+            
+        } catch (error) {
+            console.error('Error loading filters:', error);
+        }
+    }
+
+    renderClientFilter(clients) {
+        const select = document.getElementById('client-filter');
+        if (!select) return;
+        
+        // Clear existing options except "All Clients"
+        while (select.options.length > 1) {
+            select.remove(1);
+        }
+        
+        // Add client options
+        clients.forEach(client => {
+            const option = document.createElement('option');
+            option.value = client.id;
+            option.textContent = client.name;
+            select.appendChild(option);
+        });
+    }
+
+    async loadCharts() {
+        await this.loadAnalyticsChart();
+        await this.loadClientsChart();
+        await this.loadModelsChart();
+    }
+
+    async loadAnalyticsChart() {
+        try {
+            // Generate demo data
+            const labels = window.chartManager.generateDateLabels(7);
+            const data = {
+                labels: labels,
+                datasets: [
+                    {
+                        label: 'Requests',
+                        data: labels.map(() => Math.floor(Math.random() * 1000) + 500),
+                        color: '#3b82f6',
+                        fill: true
+                    },
+                    {
+                        label: 'Tokens',
+                        data: labels.map(() => Math.floor(Math.random() * 100000) + 50000),
+                        color: '#10b981',
+                        fill: true
+                    },
+                    {
+                        label: 'Cost ($)',
+                        data: labels.map(() => Math.random() * 50 + 10),
+                        color: '#f59e0b',
+                        fill: true
+                    }
+                ]
+            };
+            
+            // Create chart
+            window.chartManager.createLineChart('analytics-chart', data, {
+                scales: {
+                    y: {
+                        ticks: {
+                            callback: function(value) {
+                                return value.toLocaleString();
+                            }
+                        }
+                    }
+                }
+            });
+            
+        } catch (error) {
+            console.error('Error loading analytics chart:', error);
+        }
+    }
+
+    async loadClientsChart() {
+        try {
+            const data = {
+                labels: ['Web App', 'Mobile App', 'API Integration', 'Internal Tools', 'Testing'],
+                datasets: [{
+                    label: 'Requests',
+                    data: [45, 25, 15, 10, 5],
+                    color: '#3b82f6'
+                }]
+            };
+            
+            window.chartManager.createHorizontalBarChart('clients-chart', data);
+            
+        } catch (error) {
+            console.error('Error loading clients chart:', error);
+        }
+    }
+
+    async loadModelsChart() {
+        try {
+            const data = {
+                labels: ['gpt-4', 'gpt-3.5-turbo', 'gemini-pro', 'deepseek-chat', 'grok-beta'],
+                data: [35, 30, 20, 10, 5],
+                colors: ['#3b82f6', '#60a5fa', '#10b981', '#f59e0b', '#8b5cf6']
+            };
+            
+            window.chartManager.createDoughnutChart('models-chart', data);
+            
+        } catch (error) {
+            console.error('Error loading models chart:', error);
+        }
+    }
+
+    async loadUsageData() {
+        try {
+            // In a real app, this would fetch from /api/usage/detailed
+            const usageData = [
+                { date: '2024-01-15', client: 'Web App', provider: 'OpenAI', model: 'gpt-4', requests: 245, tokens: 125000, cost: 12.50 },
+                { date: '2024-01-15', client: 'Mobile App', provider: 'Gemini', model: 'gemini-pro', requests: 180, tokens: 89000, cost: 8.90 },
+                { date: '2024-01-15', client: 'API Integration', provider: 'OpenAI', model: 'gpt-3.5-turbo', requests: 320, tokens: 156000, cost: 15.60 },
+                { date: '2024-01-14', client: 'Web App', provider: 'OpenAI', model: 'gpt-4', requests: 210, tokens: 110000, cost: 11.00 },
+                { date: '2024-01-14', client: 'Internal Tools', provider: 'DeepSeek', model: 'deepseek-chat', requests: 95, tokens: 48000, cost: 4.80 },
+                { date: '2024-01-14', client: 'Testing Suite', provider: 'Grok', model: 'grok-beta', requests: 45, tokens: 22000, cost: 2.20 },
+                { date: '2024-01-13', client: 'Web App', provider: 'OpenAI', model: 'gpt-4', requests: 195, tokens: 98000, cost: 9.80 },
+                { date: '2024-01-13', client: 'Mobile App', provider: 'Gemini', model: 'gemini-pro', requests: 165, tokens: 82000, cost: 8.20 },
+                { date: '2024-01-13', client: 'API Integration', provider: 'OpenAI', model: 'gpt-3.5-turbo', requests: 285, tokens: 142000, cost: 14.20 },
+                { date: '2024-01-12', client: 'Web App', provider: 'OpenAI', model: 'gpt-4', requests: 230, tokens: 118000, cost: 11.80 }
+            ];
+            
+            this.renderUsageTable(usageData);
+            
+        } catch (error) {
+            console.error('Error loading usage data:', error);
+        }
+    }
+
+    renderUsageTable(data) {
+        const tableBody = document.querySelector('#usage-table tbody');
+        if (!tableBody) return;
+        
+        tableBody.innerHTML = data.map(row => `
+            <tr>
+                <td>${row.date}</td>
+                <td>${row.client}</td>
+                <td>${row.provider}</td>
+                <td>${row.model}</td>
+                <td>${row.requests.toLocaleString()}</td>
+                <td>${row.tokens.toLocaleString()}</td>
+                <td>$${row.cost.toFixed(2)}</td>
+            </tr>
+        `).join('');
+    }
+
+    setupEventListeners() {
+        // Filter controls
+        const dateRangeSelect = document.getElementById('date-range');
+        const clientSelect = document.getElementById('client-filter');
+        const providerSelect = document.getElementById('provider-filter');
+        
+        if (dateRangeSelect) {
+            dateRangeSelect.addEventListener('change', (e) => {
+                this.filters.dateRange = e.target.value;
+                this.applyFilters();
+            });
+        }
+        
+        if (clientSelect) {
+            clientSelect.addEventListener('change', (e) => {
+                this.filters.client = e.target.value;
+                this.applyFilters();
+            });
+        }
+        
+        if (providerSelect) {
+            providerSelect.addEventListener('change', (e) => {
+                this.filters.provider = e.target.value;
+                this.applyFilters();
+            });
+        }
+        
+        // Chart metric buttons
+        const metricButtons = document.querySelectorAll('.chart-control-btn[data-metric]');
+        metricButtons.forEach(button => {
+            button.addEventListener('click', () => {
+                // Update active state
+                metricButtons.forEach(btn => btn.classList.remove('active'));
+                button.classList.add('active');
+                
+                // Update chart based on metric
+                this.updateAnalyticsChart(button.dataset.metric);
+            });
+        });
+        
+        // Export button
+        const exportBtn = document.querySelector('#analytics .btn-secondary');
+        if (exportBtn) {
+            exportBtn.addEventListener('click', () => {
+                this.exportData();
+            });
+        }
+    }
+
+    applyFilters() {
+        console.log('Applying filters:', this.filters);
+        // In a real app, this would fetch filtered data from the API
+        // For now, just show a toast
+        if (window.authManager) {
+            window.authManager.showToast('Filters applied', 'success');
+        }
+        
+        // Refresh data
+        this.loadCharts();
+        this.loadUsageData();
+    }
+
+    updateAnalyticsChart(metric) {
+        // Update the main analytics chart to show the selected metric
+        const labels = window.chartManager.generateDateLabels(7);
+        
+        let data;
+        if (metric === 'requests') {
+            data = {
+                labels: labels,
+                datasets: [{
+                    label: 'Requests',
+                    data: labels.map(() => Math.floor(Math.random() * 1000) + 500),
+                    color: '#3b82f6',
+                    fill: true
+                }]
+            };
+        } else if (metric === 'tokens') {
+            data = {
+                labels: labels,
+                datasets: [{
+                    label: 'Tokens',
+                    data: labels.map(() => Math.floor(Math.random() * 100000) + 50000),
+                    color: '#10b981',
+                    fill: true
+                }]
+            };
+        } else if (metric === 'cost') {
+            data = {
+                labels: labels,
+                datasets: [{
+                    label: 'Cost ($)',
+                    data: labels.map(() => Math.random() * 50 + 10),
+                    color: '#f59e0b',
+                    fill: true
+                }]
+            };
+        }
+        
+        window.chartManager.updateChartData('analytics-chart', data);
+    }
+
+    exportData() {
+        // Create CSV data
+        const table = document.getElementById('usage-table');
+        if (!table) return;
+        
+        const rows = table.querySelectorAll('tr');
+        const csv = [];
+        
+        rows.forEach(row => {
+            const rowData = [];
+            row.querySelectorAll('th, td').forEach(cell => {
+                rowData.push(`"${cell.textContent.replace(/"/g, '""')}"`);
+            });
+            csv.push(rowData.join(','));
+        });
+        
+        // Create download link
+        const blob = new Blob([csv.join('\n')], { type: 'text/csv' });
+        const url = URL.createObjectURL(blob);
+        const a = document.createElement('a');
+        a.href = url;
+        a.download = `llm-proxy-analytics-${new Date().toISOString().split('T')[0]}.csv`;
+        document.body.appendChild(a);
+        a.click();
+        document.body.removeChild(a);
+        URL.revokeObjectURL(url);
+        
+        // Show success message
+        if (window.authManager) {
+            window.authManager.showToast('Data exported successfully', 'success');
+        }
+    }
+
+    refresh() {
+        this.loadCharts();
+        this.loadUsageData();
+    }
+}
+
+// Initialize analytics page when needed
+window.initAnalytics = async () => {
+    window.analyticsPage = new AnalyticsPage();
+};
+
+// Export for use in other modules
+if (typeof module !== 'undefined' && module.exports) {
+    module.exports = AnalyticsPage;
+}

static/js/pages/clients.js

@@ -0,0 +1,471 @@
+// Clients Page Module
+
+class ClientsPage {
+    constructor() {
+        this.clients = [];
+        this.init();
+    }
+
+    async init() {
+        // Load data
+        await this.loadClients();
+        await this.loadClientUsageChart();
+        await this.loadRateLimitStatus();
+        
+        // Setup event listeners
+        this.setupEventListeners();
+    }
+
+    async loadClients() {
+        try {
+            // In a real app, this would fetch from /api/clients
+            this.clients = [
+                { id: 'client-1', name: 'Web Application', token: 'sk-*****abc123', created: '2024-01-01', lastUsed: '2024-01-15', requests: 1245, status: 'active' },
+                { id: 'client-2', name: 'Mobile App', token: 'sk-*****def456', created: '2024-01-05', lastUsed: '2024-01-15', requests: 890, status: 'active' },
+                { id: 'client-3', name: 'API Integration', token: 'sk-*****ghi789', created: '2024-01-08', lastUsed: '2024-01-14', requests: 1560, status: 'active' },
+                { id: 'client-4', name: 'Internal Tools', token: 'sk-*****jkl012', created: '2024-01-10', lastUsed: '2024-01-13', requests: 340, status: 'inactive' },
+                { id: 'client-5', name: 'Testing Suite', token: 'sk-*****mno345', created: '2024-01-12', lastUsed: '2024-01-12', requests: 120, status: 'active' },
+                { id: 'client-6', name: 'Backup Service', token: 'sk-*****pqr678', created: '2024-01-14', lastUsed: null, requests: 0, status: 'pending' }
+            ];
+            
+            this.renderClientsTable();
+            
+        } catch (error) {
+            console.error('Error loading clients:', error);
+        }
+    }
+
+    renderClientsTable() {
+        const tableBody = document.querySelector('#clients-table tbody');
+        if (!tableBody) return;
+        
+        tableBody.innerHTML = this.clients.map(client => {
+            const statusClass = client.status === 'active' ? 'success' : 
+                              client.status === 'inactive' ? 'warning' : 'secondary';
+            const statusIcon = client.status === 'active' ? 'check-circle' :
+                              client.status === 'inactive' ? 'exclamation-triangle' : 'clock';
+            
+            return `
+                <tr>
+                    <td>${client.id}</td>
+                    <td>${client.name}</td>
+                    <td>
+                        <code class="token-display">${client.token}</code>
+                        <button class="btn-copy-token" data-token="${client.token}" title="Copy token">
+                            <i class="fas fa-copy"></i>
+                        </button>
+                    </td>
+                    <td>${client.created}</td>
+                    <td>${client.lastUsed || 'Never'}</td>
+                    <td>${client.requests.toLocaleString()}</td>
+                    <td>
+                        <span class="status-badge ${statusClass}">
+                            <i class="fas fa-${statusIcon}"></i>
+                            ${client.status}
+                        </span>
+                    </td>
+                    <td>
+                        <div class="action-buttons">
+                            <button class="btn-action" title="Edit" data-action="edit" data-id="${client.id}">
+                                <i class="fas fa-edit"></i>
+                            </button>
+                            <button class="btn-action" title="Rotate Token" data-action="rotate" data-id="${client.id}">
+                                <i class="fas fa-redo"></i>
+                            </button>
+                            <button class="btn-action danger" title="Revoke" data-action="revoke" data-id="${client.id}">
+                                <i class="fas fa-trash"></i>
+                            </button>
+                        </div>
+                    </td>
+                </tr>
+            `;
+        }).join('');
+        
+        // Add CSS for action buttons
+        this.addActionStyles();
+    }
+
+    addActionStyles() {
+        const style = document.createElement('style');
+        style.textContent = `
+            .token-display {
+                background-color: var(--bg-secondary);
+                padding: 0.25rem 0.5rem;
+                border-radius: 4px;
+                font-family: monospace;
+                font-size: 0.75rem;
+                margin-right: 0.5rem;
+            }
+            
+            .btn-copy-token {
+                background: none;
+                border: none;
+                color: var(--text-secondary);
+                cursor: pointer;
+                font-size: 0.875rem;
+                padding: 0.25rem;
+                transition: color 0.2s ease;
+            }
+            
+            .btn-copy-token:hover {
+                color: var(--primary);
+            }
+            
+            .action-buttons {
+                display: flex;
+                gap: 0.5rem;
+            }
+            
+            .btn-action {
+                background: none;
+                border: none;
+                color: var(--text-secondary);
+                cursor: pointer;
+                font-size: 0.875rem;
+                padding: 0.25rem;
+                transition: color 0.2s ease;
+            }
+            
+            .btn-action:hover {
+                color: var(--primary);
+            }
+            
+            .btn-action.danger:hover {
+                color: var(--danger);
+            }
+        `;
+        document.head.appendChild(style);
+    }
+
+    async loadClientUsageChart() {
+        try {
+            const data = {
+                labels: ['Web App', 'Mobile App', 'API Integration', 'Internal Tools', 'Testing'],
+                datasets: [{
+                    label: 'Requests',
+                    data: [1245, 890, 1560, 340, 120],
+                    color: '#3b82f6'
+                }]
+            };
+            
+            window.chartManager.createHorizontalBarChart('client-usage-chart', data);
+            
+        } catch (error) {
+            console.error('Error loading client usage chart:', error);
+        }
+    }
+
+    async loadRateLimitStatus() {
+        const container = document.getElementById('rate-limit-status');
+        if (!container) return;
+        
+        const rateLimits = [
+            { client: 'Web Application', limit: 1000, used: 645, remaining: 355 },
+            { client: 'Mobile App', limit: 500, used: 320, remaining: 180 },
+            { client: 'API Integration', limit: 2000, used: 1560, remaining: 440 },
+            { client: 'Internal Tools', limit: 100, used: 34, remaining: 66 },
+            { client: 'Testing Suite', limit: 200, used: 120, remaining: 80 }
+        ];
+        
+        container.innerHTML = rateLimits.map(limit => {
+            const percentage = (limit.used / limit.limit) * 100;
+            let color = 'success';
+            if (percentage > 80) color = 'warning';
+            if (percentage > 95) color = 'danger';
+            
+            return `
+                <div class="rate-limit-item">
+                    <div class="rate-limit-header">
+                        <span class="rate-limit-client">${limit.client}</span>
+                        <span class="rate-limit-numbers">${limit.used} / ${limit.limit}</span>
+                    </div>
+                    <div class="progress-bar">
+                        <div class="progress-fill ${color}" style="width: ${percentage}%"></div>
+                    </div>
+                    <div class="rate-limit-footer">
+                        <span class="rate-limit-percentage">${Math.round(percentage)}% used</span>
+                        <span class="rate-limit-remaining">${limit.remaining} remaining</span>
+                    </div>
+                </div>
+            `;
+        }).join('');
+        
+        // Add CSS for rate limit items
+        this.addRateLimitStyles();
+    }
+
+    addRateLimitStyles() {
+        const style = document.createElement('style');
+        style.textContent = `
+            .rate-limit-item {
+                margin-bottom: 1rem;
+            }
+            
+            .rate-limit-header {
+                display: flex;
+                justify-content: space-between;
+                align-items: center;
+                margin-bottom: 0.5rem;
+            }
+            
+            .rate-limit-client {
+                font-size: 0.875rem;
+                color: var(--text-primary);
+            }
+            
+            .rate-limit-numbers {
+                font-size: 0.875rem;
+                color: var(--text-secondary);
+            }
+            
+            .rate-limit-footer {
+                display: flex;
+                justify-content: space-between;
+                align-items: center;
+                margin-top: 0.5rem;
+                font-size: 0.75rem;
+            }
+            
+            .rate-limit-percentage {
+                color: var(--text-secondary);
+            }
+            
+            .rate-limit-remaining {
+                color: var(--success);
+                font-weight: 500;
+            }
+        `;
+        document.head.appendChild(style);
+    }
+
+    setupEventListeners() {
+        // Add client button
+        const addBtn = document.getElementById('add-client');
+        if (addBtn) {
+            addBtn.addEventListener('click', () => {
+                this.showAddClientModal();
+            });
+        }
+        
+        // Copy token buttons
+        document.addEventListener('click', (e) => {
+            if (e.target.closest('.btn-copy-token')) {
+                const button = e.target.closest('.btn-copy-token');
+                const token = button.dataset.token;
+                this.copyToClipboard(token);
+                
+                if (window.authManager) {
+                    window.authManager.showToast('Token copied to clipboard', 'success');
+                }
+            }
+        });
+        
+        // Action buttons
+        document.addEventListener('click', (e) => {
+            if (e.target.closest('.btn-action')) {
+                const button = e.target.closest('.btn-action');
+                const action = button.dataset.action;
+                const clientId = button.dataset.id;
+                
+                switch (action) {
+                    case 'edit':
+                        this.editClient(clientId);
+                        break;
+                    case 'rotate':
+                        this.rotateToken(clientId);
+                        break;
+                    case 'revoke':
+                        this.revokeClient(clientId);
+                        break;
+                }
+            }
+        });
+    }
+
+    showAddClientModal() {
+        const modal = document.createElement('div');
+        modal.className = 'modal active';
+        modal.innerHTML = `
+            <div class="modal-content">
+                <div class="modal-header">
+                    <h3 class="modal-title">Add New Client</h3>
+                    <button class="modal-close">
+                        <i class="fas fa-times"></i>
+                    </button>
+                </div>
+                <div class="modal-body">
+                    <form id="add-client-form">
+                        <div class="form-control">
+                            <label for="client-name">Client Name</label>
+                            <input type="text" id="client-name" placeholder="e.g., Web Application" required>
+                        </div>
+                        <div class="form-control">
+                            <label for="client-description">Description (Optional)</label>
+                            <textarea id="client-description" rows="3" placeholder="Describe what this client will be used for..."></textarea>
+                        </div>
+                        <div class="form-control">
+                            <label for="rate-limit">Rate Limit (requests per hour)</label>
+                            <input type="number" id="rate-limit" value="1000" min="1" max="10000">
+                        </div>
+                    </form>
+                </div>
+                <div class="modal-footer">
+                    <button class="btn btn-secondary close-modal">Cancel</button>
+                    <button class="btn btn-primary create-client">Create Client</button>
+                </div>
+            </div>
+        `;
+        
+        document.body.appendChild(modal);
+        
+        // Setup event listeners
+        const closeBtn = modal.querySelector('.modal-close');
+        const closeModalBtn = modal.querySelector('.close-modal');
+        const createBtn = modal.querySelector('.create-client');
+        
+        const closeModal = () => {
+            modal.classList.remove('active');
+            setTimeout(() => modal.remove(), 300);
+        };
+        
+        closeBtn.addEventListener('click', closeModal);
+        closeModalBtn.addEventListener('click', closeModal);
+        
+        createBtn.addEventListener('click', () => {
+            const name = modal.querySelector('#client-name').value;
+            if (!name.trim()) {
+                if (window.authManager) {
+                    window.authManager.showToast('Client name is required', 'error');
+                }
+                return;
+            }
+            
+            // In a real app, this would create the client via API
+            if (window.authManager) {
+                window.authManager.showToast(`Client "${name}" created successfully`, 'success');
+            }
+            
+            // Refresh clients list
+            this.loadClients();
+            closeModal();
+        });
+        
+        // Close on background click
+        modal.addEventListener('click', (e) => {
+            if (e.target === modal) {
+                closeModal();
+            }
+        });
+    }
+
+    editClient(clientId) {
+        const client = this.clients.find(c => c.id === clientId);
+        if (!client) return;
+        
+        // Show edit modal
+        const modal = document.createElement('div');
+        modal.className = 'modal active';
+        modal.innerHTML = `
+            <div class="modal-content">
+                <div class="modal-header">
+                    <h3 class="modal-title">Edit Client: ${client.name}</h3>
+                    <button class="modal-close">
+                        <i class="fas fa-times"></i>
+                    </button>
+                </div>
+                <div class="modal-body">
+                    <p>Client editing would be implemented here.</p>
+                    <p>In a real implementation, this would include forms for updating client settings.</p>
+                </div>
+                <div class="modal-footer">
+                    <button class="btn btn-secondary close-modal">Cancel</button>
+                    <button class="btn btn-primary save-client">Save Changes</button>
+                </div>
+            </div>
+        `;
+        
+        document.body.appendChild(modal);
+        
+        // Setup event listeners
+        const closeBtn = modal.querySelector('.modal-close');
+        const closeModalBtn = modal.querySelector('.close-modal');
+        const saveBtn = modal.querySelector('.save-client');
+        
+        const closeModal = () => {
+            modal.classList.remove('active');
+            setTimeout(() => modal.remove(), 300);
+        };
+        
+        closeBtn.addEventListener('click', closeModal);
+        closeModalBtn.addEventListener('click', closeModal);
+        
+        saveBtn.addEventListener('click', () => {
+            // In a real app, this would save client changes
+            if (window.authManager) {
+                window.authManager.showToast('Client updated successfully', 'success');
+            }
+            closeModal();
+        });
+        
+        // Close on background click
+        modal.addEventListener('click', (e) => {
+            if (e.target === modal) {
+                closeModal();
+            }
+        });
+    }
+
+    rotateToken(clientId) {
+        const client = this.clients.find(c => c.id === clientId);
+        if (!client) return;
+        
+        // Show confirmation modal
+        if (confirm(`Are you sure you want to rotate the token for "${client.name}"? The old token will be invalidated.`)) {
+            // In a real app, this would rotate the token via API
+            if (window.authManager) {
+                window.authManager.showToast(`Token rotated for "${client.name}"`, 'success');
+            }
+            
+            // Refresh clients list
+            this.loadClients();
+        }
+    }
+
+    revokeClient(clientId) {
+        const client = this.clients.find(c => c.id === clientId);
+        if (!client) return;
+        
+        // Show confirmation modal
+        if (confirm(`Are you sure you want to revoke client "${client.name}"? This action cannot be undone.`)) {
+            // In a real app, this would revoke the client via API
+            if (window.authManager) {
+                window.authManager.showToast(`Client "${client.name}" revoked`, 'success');
+            }
+            
+            // Refresh clients list
+            this.loadClients();
+        }
+    }
+
+    copyToClipboard(text) {
+        navigator.clipboard.writeText(text).catch(err => {
+            console.error('Failed to copy:', err);
+        });
+    }
+
+    refresh() {
+        this.loadClients();
+        this.loadClientUsageChart();
+        this.loadRateLimitStatus();
+    }
+}
+
+// Initialize clients page when needed
+window.initClients = async () => {
+    window.clientsPage = new ClientsPage();
+};
+
+// Export for use in other modules
+if (typeof module !== 'undefined' && module.exports) {
+    module.exports = ClientsPage;
+}

static/js/pages/costs.js

@@ -0,0 +1,468 @@
+// Costs Page Module
+
+class CostsPage {
+    constructor() {
+        this.costData = null;
+        this.init();
+    }
+
+    async init() {
+        // Load data
+        await this.loadCostStats();
+        await this.loadCostsChart();
+        await this.loadBudgetTracking();
+        await this.loadCostProjections();
+        await this.loadPricingTable();
+        
+        // Setup event listeners
+        this.setupEventListeners();
+    }
+
+    async loadCostStats() {
+        try {
+            // In a real app, this would fetch from /api/costs/summary
+            this.costData = {
+                totalCost: 125.43,
+                todayCost: 12.45,
+                weekCost: 45.67,
+                monthCost: 125.43,
+                avgDailyCost: 8.36,
+                costTrend: 5.2, // percentage
+                budgetUsed: 62, // percentage
+                projectedMonthEnd: 189.75
+            };
+            
+            this.renderCostStats();
+            
+        } catch (error) {
+            console.error('Error loading cost stats:', error);
+        }
+    }
+
+    renderCostStats() {
+        const container = document.getElementById('cost-stats');
+        if (!container) return;
+        
+        container.innerHTML = `
+            <div class="stat-card">
+                <div class="stat-icon warning">
+                    <i class="fas fa-dollar-sign"></i>
+                </div>
+                <div class="stat-content">
+                    <div class="stat-value">$${this.costData.totalCost.toFixed(2)}</div>
+                    <div class="stat-label">Total Cost</div>
+                    <div class="stat-change positive">
+                        <i class="fas fa-arrow-up"></i>
+                        $${this.costData.todayCost.toFixed(2)} today
+                    </div>
+                </div>
+            </div>
+            
+            <div class="stat-card">
+                <div class="stat-icon primary">
+                    <i class="fas fa-calendar-week"></i>
+                </div>
+                <div class="stat-content">
+                    <div class="stat-value">$${this.costData.weekCost.toFixed(2)}</div>
+                    <div class="stat-label">This Week</div>
+                    <div class="stat-change ${this.costData.costTrend > 0 ? 'positive' : 'negative'}">
+                        <i class="fas fa-arrow-${this.costData.costTrend > 0 ? 'up' : 'down'}"></i>
+                        ${Math.abs(this.costData.costTrend)}% from last week
+                    </div>
+                </div>
+            </div>
+            
+            <div class="stat-card">
+                <div class="stat-icon success">
+                    <i class="fas fa-calendar-alt"></i>
+                </div>
+                <div class="stat-content">
+                    <div class="stat-value">$${this.costData.monthCost.toFixed(2)}</div>
+                    <div class="stat-label">This Month</div>
+                    <div class="stat-change">
+                        <i class="fas fa-chart-line"></i>
+                        $${this.costData.avgDailyCost.toFixed(2)}/day avg
+                    </div>
+                </div>
+            </div>
+            
+            <div class="stat-card">
+                <div class="stat-icon danger">
+                    <i class="fas fa-piggy-bank"></i>
+                </div>
+                <div class="stat-content">
+                    <div class="stat-value">${this.costData.budgetUsed}%</div>
+                    <div class="stat-label">Budget Used</div>
+                    <div class="stat-change">
+                        <i class="fas fa-project-diagram"></i>
+                        $${this.costData.projectedMonthEnd.toFixed(2)} projected
+                    </div>
+                </div>
+            </div>
+        `;
+    }
+
+    async loadCostsChart() {
+        try {
+            // Generate demo data
+            const data = {
+                labels: ['OpenAI', 'Gemini', 'DeepSeek', 'Grok'],
+                datasets: [{
+                    label: 'Cost by Provider',
+                    data: [65, 25, 8, 2],
+                    color: '#3b82f6'
+                }]
+            };
+            
+            window.chartManager.createBarChart('costs-chart', data, {
+                plugins: {
+                    tooltip: {
+                        callbacks: {
+                            label: function(context) {
+                                return `$${context.parsed.y.toFixed(2)} (${context.parsed.y}%)`;
+                            }
+                        }
+                    }
+                }
+            });
+            
+        } catch (error) {
+            console.error('Error loading costs chart:', error);
+        }
+    }
+
+    async loadBudgetTracking() {
+        const container = document.getElementById('budget-progress');
+        if (!container) return;
+        
+        const budgets = [
+            { name: 'Monthly Budget', used: 62, total: 200, color: 'primary' },
+            { name: 'OpenAI Budget', used: 75, total: 150, color: 'info' },
+            { name: 'Gemini Budget', used: 45, total: 100, color: 'success' },
+            { name: 'Team Budget', used: 30, total: 50, color: 'warning' }
+        ];
+        
+        container.innerHTML = budgets.map(budget => `
+            <div class="budget-item">
+                <div class="budget-header">
+                    <span class="budget-name">${budget.name}</span>
+                    <span class="budget-amount">$${budget.used} / $${budget.total}</span>
+                </div>
+                <div class="progress-bar">
+                    <div class="progress-fill ${budget.color}" style="width: ${(budget.used / budget.total) * 100}%"></div>
+                </div>
+                <div class="budget-footer">
+                    <span class="budget-percentage">${Math.round((budget.used / budget.total) * 100)}% used</span>
+                    <span class="budget-remaining">$${budget.total - budget.used} remaining</span>
+                </div>
+            </div>
+        `).join('');
+        
+        // Add CSS for budget items
+        this.addBudgetStyles();
+    }
+
+    addBudgetStyles() {
+        const style = document.createElement('style');
+        style.textContent = `
+            .budget-item {
+                margin-bottom: 1.5rem;
+            }
+            
+            .budget-header {
+                display: flex;
+                justify-content: space-between;
+                align-items: center;
+                margin-bottom: 0.5rem;
+            }
+            
+            .budget-name {
+                font-size: 0.875rem;
+                font-weight: 500;
+                color: var(--text-primary);
+            }
+            
+            .budget-amount {
+                font-size: 0.875rem;
+                color: var(--text-secondary);
+            }
+            
+            .budget-footer {
+                display: flex;
+                justify-content: space-between;
+                align-items: center;
+                margin-top: 0.5rem;
+                font-size: 0.75rem;
+            }
+            
+            .budget-percentage {
+                color: var(--text-secondary);
+            }
+            
+            .budget-remaining {
+                color: var(--success);
+                font-weight: 500;
+            }
+            
+            .progress-fill.primary {
+                background-color: var(--primary);
+            }
+            
+            .progress-fill.info {
+                background-color: var(--info);
+            }
+            
+            .progress-fill.success {
+                background-color: var(--success);
+            }
+            
+            .progress-fill.warning {
+                background-color: var(--warning);
+            }
+        `;
+        document.head.appendChild(style);
+    }
+
+    async loadCostProjections() {
+        const container = document.getElementById('cost-projections');
+        if (!container) return;
+        
+        const projections = [
+            { period: 'Today', amount: 12.45, trend: 'up' },
+            { period: 'This Week', amount: 45.67, trend: 'up' },
+            { period: 'This Month', amount: 189.75, trend: 'up' },
+            { period: 'Next Month', amount: 210.50, trend: 'up' }
+        ];
+        
+        container.innerHTML = projections.map(proj => `
+            <div class="projection-item">
+                <div class="projection-period">${proj.period}</div>
+                <div class="projection-amount">$${proj.amount.toFixed(2)}</div>
+                <div class="projection-trend ${proj.trend}">
+                    <i class="fas fa-arrow-${proj.trend}"></i>
+                </div>
+            </div>
+        `).join('');
+        
+        // Add CSS for projections
+        this.addProjectionStyles();
+    }
+
+    addProjectionStyles() {
+        const style = document.createElement('style');
+        style.textContent = `
+            .projection-item {
+                display: flex;
+                align-items: center;
+                justify-content: space-between;
+                padding: 0.75rem 0;
+                border-bottom: 1px solid var(--border-color);
+            }
+            
+            .projection-item:last-child {
+                border-bottom: none;
+            }
+            
+            .projection-period {
+                font-size: 0.875rem;
+                color: var(--text-primary);
+            }
+            
+            .projection-amount {
+                font-size: 1rem;
+                font-weight: 600;
+                color: var(--text-primary);
+            }
+            
+            .projection-trend {
+                width: 24px;
+                height: 24px;
+                border-radius: 50%;
+                display: flex;
+                align-items: center;
+                justify-content: center;
+            }
+            
+            .projection-trend.up {
+                background-color: rgba(239, 68, 68, 0.1);
+                color: var(--danger);
+            }
+            
+            .projection-trend.down {
+                background-color: rgba(16, 185, 129, 0.1);
+                color: var(--success);
+            }
+        `;
+        document.head.appendChild(style);
+    }
+
+    async loadPricingTable() {
+        try {
+            // In a real app, this would fetch from /api/pricing
+            const pricingData = [
+                { provider: 'OpenAI', model: 'gpt-4', input: 0.03, output: 0.06, updated: '2024-01-15' },
+                { provider: 'OpenAI', model: 'gpt-3.5-turbo', input: 0.0015, output: 0.002, updated: '2024-01-15' },
+                { provider: 'Gemini', model: 'gemini-pro', input: 0.0005, output: 0.0015, updated: '2024-01-14' },
+                { provider: 'Gemini', model: 'gemini-pro-vision', input: 0.0025, output: 0.0075, updated: '2024-01-14' },
+                { provider: 'DeepSeek', model: 'deepseek-chat', input: 0.00014, output: 0.00028, updated: '2024-01-13' },
+                { provider: 'DeepSeek', model: 'deepseek-coder', input: 0.00014, output: 0.00028, updated: '2024-01-13' },
+                { provider: 'Grok', model: 'grok-beta', input: 0.01, output: 0.03, updated: '2024-01-12' }
+            ];
+            
+            this.renderPricingTable(pricingData);
+            
+        } catch (error) {
+            console.error('Error loading pricing data:', error);
+        }
+    }
+
+    renderPricingTable(data) {
+        const tableBody = document.querySelector('#pricing-table tbody');
+        if (!tableBody) return;
+        
+        tableBody.innerHTML = data.map(row => `
+            <tr>
+                <td>${row.provider}</td>
+                <td>${row.model}</td>
+                <td>$${row.input.toFixed(5)}/1K tokens</td>
+                <td>$${row.output.toFixed(5)}/1K tokens</td>
+                <td>${row.updated}</td>
+            </tr>
+        `).join('');
+    }
+
+    setupEventListeners() {
+        // Breakdown buttons
+        const breakdownButtons = document.querySelectorAll('.chart-control-btn[data-breakdown]');
+        breakdownButtons.forEach(button => {
+            button.addEventListener('click', () => {
+                // Update active state
+                breakdownButtons.forEach(btn => btn.classList.remove('active'));
+                button.classList.add('active');
+                
+                // Update chart based on breakdown
+                this.updateCostsChart(button.dataset.breakdown);
+            });
+        });
+        
+        // Edit pricing button
+        const editBtn = document.getElementById('edit-pricing');
+        if (editBtn) {
+            editBtn.addEventListener('click', () => {
+                this.editPricing();
+            });
+        }
+    }
+
+    updateCostsChart(breakdown) {
+        let data;
+        
+        if (breakdown === 'provider') {
+            data = {
+                labels: ['OpenAI', 'Gemini', 'DeepSeek', 'Grok'],
+                datasets: [{
+                    label: 'Cost by Provider',
+                    data: [65, 25, 8, 2],
+                    color: '#3b82f6'
+                }]
+            };
+        } else if (breakdown === 'client') {
+            data = {
+                labels: ['Web App', 'Mobile App', 'API Integration', 'Internal Tools', 'Testing'],
+                datasets: [{
+                    label: 'Cost by Client',
+                    data: [40, 25, 20, 10, 5],
+                    color: '#10b981'
+                }]
+            };
+        } else if (breakdown === 'model') {
+            data = {
+                labels: ['gpt-4', 'gpt-3.5-turbo', 'gemini-pro', 'deepseek-chat', 'grok-beta'],
+                datasets: [{
+                    label: 'Cost by Model',
+                    data: [35, 30, 20, 10, 5],
+                    color: '#f59e0b'
+                }]
+            };
+        }
+        
+        window.chartManager.updateChartData('costs-chart', data);
+    }
+
+    editPricing() {
+        // Show pricing edit modal
+        this.showPricingModal();
+    }
+
+    showPricingModal() {
+        // Create modal for editing pricing
+        const modal = document.createElement('div');
+        modal.className = 'modal active';
+        modal.innerHTML = `
+            <div class="modal-content">
+                <div class="modal-header">
+                    <h3 class="modal-title">Edit Pricing Configuration</h3>
+                    <button class="modal-close">
+                        <i class="fas fa-times"></i>
+                    </button>
+                </div>
+                <div class="modal-body">
+                    <p>Pricing configuration would be editable here.</p>
+                    <p>In a real implementation, this would include forms for updating provider pricing.</p>
+                </div>
+                <div class="modal-footer">
+                    <button class="btn btn-secondary close-modal">Cancel</button>
+                    <button class="btn btn-primary save-pricing">Save Changes</button>
+                </div>
+            </div>
+        `;
+        
+        document.body.appendChild(modal);
+        
+        // Setup event listeners
+        const closeBtn = modal.querySelector('.modal-close');
+        const closeModalBtn = modal.querySelector('.close-modal');
+        const saveBtn = modal.querySelector('.save-pricing');
+        
+        const closeModal = () => {
+            modal.classList.remove('active');
+            setTimeout(() => modal.remove(), 300);
+        };
+        
+        closeBtn.addEventListener('click', closeModal);
+        closeModalBtn.addEventListener('click', closeModal);
+        
+        saveBtn.addEventListener('click', () => {
+            // In a real app, this would save pricing changes
+            if (window.authManager) {
+                window.authManager.showToast('Pricing updated successfully', 'success');
+            }
+            closeModal();
+        });
+        
+        // Close on background click
+        modal.addEventListener('click', (e) => {
+            if (e.target === modal) {
+                closeModal();
+            }
+        });
+    }
+
+    refresh() {
+        this.loadCostStats();
+        this.loadCostsChart();
+        this.loadBudgetTracking();
+        this.loadCostProjections();
+        this.loadPricingTable();
+    }
+}
+
+// Initialize costs page when needed
+window.initCosts = async () => {
+    window.costsPage = new CostsPage();
+};
+
+// Export for use in other modules
+if (typeof module !== 'undefined' && module.exports) {
+    module.exports = CostsPage;
+}

static/js/pages/logs.js

@@ -0,0 +1,567 @@
+// Logs Page Module
+
+class LogsPage {
+    constructor() {
+        this.logs = [];
+        this.filters = {
+            level: 'all',
+            timeRange: '24h',
+            search: ''
+        };
+        this.init();
+    }
+
+    async init() {
+        // Load logs
+        await this.loadLogs();
+        
+        // Setup event listeners
+        this.setupEventListeners();
+        
+        // Setup WebSocket subscription for live logs
+        this.setupWebSocketSubscription();
+    }
+
+    async loadLogs() {
+        try {
+            // In a real app, this would fetch from /api/system/logs
+            // Generate demo logs
+            this.generateDemoLogs(50);
+            
+            this.applyFiltersAndRender();
+            
+        } catch (error) {
+            console.error('Error loading logs:', error);
+        }
+    }
+
+    generateDemoLogs(count) {
+        const levels = ['info', 'warn', 'error', 'debug'];
+        const sources = ['server', 'database', 'auth', 'providers', 'clients', 'api'];
+        const messages = [
+            'Request processed successfully',
+            'Cache hit for model gpt-4',
+            'Rate limit check passed',
+            'High latency detected for DeepSeek provider',
+            'API key validation failed',
+            'Database connection pool healthy',
+            'New client registered: client-7',
+            'Backup completed successfully',
+            'Memory usage above 80% threshold',
+            'Provider Grok is offline',
+            'WebSocket connection established',
+            'Authentication token expired',
+            'Cost calculation completed',
+            'Rate limit exceeded for client-2',
+            'Database query optimization needed',
+            'SSL certificate renewed',
+            'System health check passed',
+            'Error in OpenAI API response',
+            'Gemini provider rate limited',
+            'DeepSeek connection timeout'
+        ];
+        
+        this.logs = [];
+        const now = Date.now();
+        
+        for (let i = 0; i < count; i++) {
+            const level = levels[Math.floor(Math.random() * levels.length)];
+            const source = sources[Math.floor(Math.random() * sources.length)];
+            const message = messages[Math.floor(Math.random() * messages.length)];
+            
+            // Generate timestamp (spread over last 24 hours)
+            const hoursAgo = Math.random() * 24;
+            const timestamp = new Date(now - hoursAgo * 60 * 60 * 1000);
+            
+            this.logs.push({
+                id: `log-${i}`,
+                timestamp: timestamp.toISOString(),
+                level: level,
+                source: source,
+                message: message,
+                details: level === 'error' ? 'Additional error details would appear here' : null
+            });
+        }
+        
+        // Sort by timestamp (newest first)
+        this.logs.sort((a, b) => new Date(b.timestamp) - new Date(a.timestamp));
+    }
+
+    applyFiltersAndRender() {
+        let filteredLogs = [...this.logs];
+        
+        // Apply level filter
+        if (this.filters.level !== 'all') {
+            filteredLogs = filteredLogs.filter(log => log.level === this.filters.level);
+        }
+        
+        // Apply time range filter
+        const now = Date.now();
+        let timeLimit = now;
+        
+        switch (this.filters.timeRange) {
+            case '1h':
+                timeLimit = now - 60 * 60 * 1000;
+                break;
+            case '24h':
+                timeLimit = now - 24 * 60 * 60 * 1000;
+                break;
+            case '7d':
+                timeLimit = now - 7 * 24 * 60 * 60 * 1000;
+                break;
+            case '30d':
+                timeLimit = now - 30 * 24 * 60 * 60 * 1000;
+                break;
+        }
+        
+        filteredLogs = filteredLogs.filter(log => new Date(log.timestamp) >= timeLimit);
+        
+        // Apply search filter
+        if (this.filters.search) {
+            const searchLower = this.filters.search.toLowerCase();
+            filteredLogs = filteredLogs.filter(log => 
+                log.message.toLowerCase().includes(searchLower) ||
+                log.source.toLowerCase().includes(searchLower) ||
+                log.level.toLowerCase().includes(searchLower)
+            );
+        }
+        
+        this.renderLogsTable(filteredLogs);
+    }
+
+    renderLogsTable(logs) {
+        const tableBody = document.querySelector('#logs-table tbody');
+        if (!tableBody) return;
+        
+        if (logs.length === 0) {
+            tableBody.innerHTML = `
+                <tr>
+                    <td colspan="4" class="empty-table">
+                        <i class="fas fa-search"></i>
+                        <div>No logs found matching your filters</div>
+                    </td>
+                </tr>
+            `;
+            return;
+        }
+        
+        tableBody.innerHTML = logs.map(log => {
+            const time = new Date(log.timestamp).toLocaleString();
+            const levelClass = `log-${log.level}`;
+            const levelIcon = this.getLevelIcon(log.level);
+            
+            return `
+                <tr class="log-row ${levelClass}" data-log-id="${log.id}">
+                    <td>${time}</td>
+                    <td>
+                        <span class="log-level-badge ${levelClass}">
+                            <i class="fas fa-${levelIcon}"></i>
+                            ${log.level.toUpperCase()}
+                        </span>
+                    </td>
+                    <td>${log.source}</td>
+                    <td>
+                        <div class="log-message">${log.message}</div>
+                        ${log.details ? `<div class="log-details">${log.details}</div>` : ''}
+                    </td>
+                </tr>
+            `;
+        }).join('');
+        
+        // Add CSS for logs table
+        this.addLogsStyles();
+    }
+
+    getLevelIcon(level) {
+        switch (level) {
+            case 'error': return 'exclamation-circle';
+            case 'warn': return 'exclamation-triangle';
+            case 'info': return 'info-circle';
+            case 'debug': return 'bug';
+            default: return 'circle';
+        }
+    }
+
+    addLogsStyles() {
+        const style = document.createElement('style');
+        style.textContent = `
+            .log-level-badge {
+                display: inline-flex;
+                align-items: center;
+                gap: 0.25rem;
+                padding: 0.25rem 0.5rem;
+                border-radius: 12px;
+                font-size: 0.75rem;
+                font-weight: 600;
+            }
+            
+            .log-error .log-level-badge {
+                background-color: rgba(239, 68, 68, 0.1);
+                color: var(--danger);
+            }
+            
+            .log-warn .log-level-badge {
+                background-color: rgba(245, 158, 11, 0.1);
+                color: var(--warning);
+            }
+            
+            .log-info .log-level-badge {
+                background-color: rgba(6, 182, 212, 0.1);
+                color: var(--info);
+            }
+            
+            .log-debug .log-level-badge {
+                background-color: rgba(100, 116, 139, 0.1);
+                color: var(--text-secondary);
+            }
+            
+            .log-message {
+                font-size: 0.875rem;
+                color: var(--text-primary);
+            }
+            
+            .log-details {
+                font-size: 0.75rem;
+                color: var(--text-secondary);
+                margin-top: 0.25rem;
+                padding: 0.5rem;
+                background-color: var(--bg-secondary);
+                border-radius: 4px;
+                border-left: 3px solid var(--danger);
+            }
+            
+            .log-row:hover {
+                background-color: var(--bg-hover);
+            }
+            
+            .empty-table {
+                text-align: center;
+                padding: 3rem !important;
+                color: var(--text-secondary);
+            }
+            
+            .empty-table i {
+                font-size: 2rem;
+                margin-bottom: 1rem;
+                opacity: 0.5;
+            }
+            
+            .empty-table div {
+                font-size: 0.875rem;
+            }
+        `;
+        document.head.appendChild(style);
+    }
+
+    setupEventListeners() {
+        // Filter controls
+        const logFilter = document.getElementById('log-filter');
+        const timeRangeFilter = document.getElementById('log-time-range');
+        const searchInput = document.getElementById('log-search');
+        
+        if (logFilter) {
+            logFilter.addEventListener('change', (e) => {
+                this.filters.level = e.target.value;
+                this.applyFiltersAndRender();
+            });
+        }
+        
+        if (timeRangeFilter) {
+            timeRangeFilter.addEventListener('change', (e) => {
+                this.filters.timeRange = e.target.value;
+                this.applyFiltersAndRender();
+            });
+        }
+        
+        if (searchInput) {
+            let searchTimeout;
+            searchInput.addEventListener('input', (e) => {
+                clearTimeout(searchTimeout);
+                searchTimeout = setTimeout(() => {
+                    this.filters.search = e.target.value;
+                    this.applyFiltersAndRender();
+                }, 300);
+            });
+        }
+        
+        // Action buttons
+        const downloadBtn = document.getElementById('download-logs');
+        const clearBtn = document.getElementById('clear-logs');
+        
+        if (downloadBtn) {
+            downloadBtn.addEventListener('click', () => {
+                this.downloadLogs();
+            });
+        }
+        
+        if (clearBtn) {
+            clearBtn.addEventListener('click', () => {
+                this.clearLogs();
+            });
+        }
+        
+        // Log row click for details
+        document.addEventListener('click', (e) => {
+            const logRow = e.target.closest('.log-row');
+            if (logRow) {
+                this.showLogDetails(logRow.dataset.logId);
+            }
+        });
+    }
+
+    setupWebSocketSubscription() {
+        if (!window.wsManager) return;
+        
+        // Subscribe to log updates
+        window.wsManager.subscribe('logs', (log) => {
+            this.addNewLog(log);
+        });
+    }
+
+    addNewLog(log) {
+        // Add to beginning of logs array
+        this.logs.unshift({
+            id: `log-${Date.now()}`,
+            timestamp: new Date().toISOString(),
+            level: log.level || 'info',
+            source: log.source || 'unknown',
+            message: log.message || '',
+            details: log.details || null
+        });
+        
+        // Keep logs array manageable
+        if (this.logs.length > 1000) {
+            this.logs = this.logs.slice(0, 1000);
+        }
+        
+        // Apply filters and re-render
+        this.applyFiltersAndRender();
+    }
+
+    downloadLogs() {
+        // Get filtered logs
+        let filteredLogs = [...this.logs];
+        
+        // Apply current filters
+        if (this.filters.level !== 'all') {
+            filteredLogs = filteredLogs.filter(log => log.level === this.filters.level);
+        }
+        
+        // Create CSV content
+        const headers = ['Timestamp', 'Level', 'Source', 'Message', 'Details'];
+        const rows = filteredLogs.map(log => [
+            new Date(log.timestamp).toISOString(),
+            log.level,
+            log.source,
+            `"${log.message.replace(/"/g, '""')}"`,
+            log.details ? `"${log.details.replace(/"/g, '""')}"` : ''
+        ]);
+        
+        const csvContent = [
+            headers.join(','),
+            ...rows.map(row => row.join(','))
+        ].join('\n');
+        
+        // Create download link
+        const blob = new Blob([csvContent], { type: 'text/csv' });
+        const url = URL.createObjectURL(blob);
+        const a = document.createElement('a');
+        a.href = url;
+        a.download = `llm-proxy-logs-${new Date().toISOString().split('T')[0]}.csv`;
+        document.body.appendChild(a);
+        a.click();
+        document.body.removeChild(a);
+        URL.revokeObjectURL(url);
+        
+        if (window.authManager) {
+            window.authManager.showToast('Logs downloaded successfully', 'success');
+        }
+    }
+
+    clearLogs() {
+        if (confirm('Are you sure you want to clear all logs? This action cannot be undone.')) {
+            // In a real app, this would clear logs via API
+            this.logs = [];
+            this.applyFiltersAndRender();
+            
+            if (window.authManager) {
+                window.authManager.showToast('Logs cleared successfully', 'success');
+            }
+        }
+    }
+
+    showLogDetails(logId) {
+        const log = this.logs.find(l => l.id === logId);
+        if (!log) return;
+        
+        // Show log details modal
+        const modal = document.createElement('div');
+        modal.className = 'modal active';
+        modal.innerHTML = `
+            <div class="modal-content" style="max-width: 800px;">
+                <div class="modal-header">
+                    <h3 class="modal-title">Log Details</h3>
+                    <button class="modal-close">
+                        <i class="fas fa-times"></i>
+                    </button>
+                </div>
+                <div class="modal-body">
+                    <div class="log-detail-grid">
+                        <div class="detail-item">
+                            <span class="detail-label">Timestamp:</span>
+                            <span class="detail-value">${new Date(log.timestamp).toLocaleString()}</span>
+                        </div>
+                        <div class="detail-item">
+                            <span class="detail-label">Level:</span>
+                            <span class="detail-value">
+                                <span class="log-level-badge log-${log.level}">
+                                    <i class="fas fa-${this.getLevelIcon(log.level)}"></i>
+                                    ${log.level.toUpperCase()}
+                                </span>
+                            </span>
+                        </div>
+                        <div class="detail-item">
+                            <span class="detail-label">Source:</span>
+                            <span class="detail-value">${log.source}</span>
+                        </div>
+                        <div class="detail-item full-width">
+                            <span class="detail-label">Message:</span>
+                            <div class="detail-value message-box">${log.message}</div>
+                        </div>
+                        ${log.details ? `
+                            <div class="detail-item full-width">
+                                <span class="detail-label">Details:</span>
+                                <div class="detail-value details-box">${log.details}</div>
+                            </div>
+                        ` : ''}
+                        <div class="detail-item full-width">
+                            <span class="detail-label">Raw JSON:</span>
+                            <pre class="detail-value json-box">${JSON.stringify(log, null, 2)}</pre>
+                        </div>
+                    </div>
+                </div>
+                <div class="modal-footer">
+                    <button class="btn btn-secondary close-modal">Close</button>
+                    <button class="btn btn-primary copy-json" data-json='${JSON.stringify(log)}'>
+                        <i class="fas fa-copy"></i> Copy JSON
+                    </button>
+                </div>
+            </div>
+        `;
+        
+        document.body.appendChild(modal);
+        
+        // Setup event listeners
+        const closeBtn = modal.querySelector('.modal-close');
+        const closeModalBtn = modal.querySelector('.close-modal');
+        const copyBtn = modal.querySelector('.copy-json');
+        
+        const closeModal = () => {
+            modal.classList.remove('active');
+            setTimeout(() => modal.remove(), 300);
+        };
+        
+        closeBtn.addEventListener('click', closeModal);
+        closeModalBtn.addEventListener('click', closeModal);
+        
+        copyBtn.addEventListener('click', () => {
+            const json = copyBtn.dataset.json;
+            navigator.clipboard.writeText(json).then(() => {
+                if (window.authManager) {
+                    window.authManager.showToast('JSON copied to clipboard', 'success');
+                }
+            }).catch(err => {
+                console.error('Failed to copy:', err);
+            });
+        });
+        
+        // Close on background click
+        modal.addEventListener('click', (e) => {
+            if (e.target === modal) {
+                closeModal();
+            }
+        });
+        
+        // Add CSS for log details
+        this.addLogDetailStyles();
+    }
+
+    addLogDetailStyles() {
+        const style = document.createElement('style');
+        style.textContent = `
+            .log-detail-grid {
+                display: grid;
+                grid-template-columns: repeat(auto-fit, minmax(300px, 1fr));
+                gap: 1rem;
+            }
+            
+            .detail-item {
+                display: flex;
+                flex-direction: column;
+                gap: 0.25rem;
+            }
+            
+            .detail-item.full-width {
+                grid-column: 1 / -1;
+            }
+            
+            .detail-label {
+                font-size: 0.875rem;
+                font-weight: 600;
+                color: var(--text-secondary);
+            }
+            
+            .detail-value {
+                font-size: 0.875rem;
+                color: var(--text-primary);
+            }
+            
+            .message-box {
+                padding: 0.75rem;
+                background-color: var(--bg-secondary);
+                border-radius: 4px;
+                border-left: 3px solid var(--primary);
+            }
+            
+            .details-box {
+                padding: 0.75rem;
+                background-color: var(--bg-secondary);
+                border-radius: 4px;
+                border-left: 3px solid var(--warning);
+                white-space: pre-wrap;
+                font-family: monospace;
+                font-size: 0.75rem;
+            }
+            
+            .json-box {
+                padding: 0.75rem;
+                background-color: #1e293b;
+                color: #e2e8f0;
+                border-radius: 4px;
+                overflow: auto;
+                max-height: 300px;
+                font-size: 0.75rem;
+                line-height: 1.5;
+            }
+        `;
+        document.head.appendChild(style);
+    }
+
+    refresh() {
+        this.loadLogs();
+        
+        if (window.authManager) {
+            window.authManager.showToast('Logs refreshed', 'success');
+        }
+    }
+}
+
+// Initialize logs page when needed
+window.initLogs = async () => {
+    window.logsPage = new LogsPage();
+};
+
+// Export for use in other modules
+if (typeof module !== 'undefined' && module.exports) {
+    module.exports = LogsPage;
+}

static/js/pages/monitoring.js

@@ -0,0 +1,611 @@
+// Monitoring Page Module
+
+class MonitoringPage {
+    constructor() {
+        this.isPaused = false;
+        this.requestStream = [];
+        this.systemLogs = [];
+        this.init();
+    }
+
+    async init() {
+        // Load initial data
+        await this.loadSystemMetrics();
+        await this.loadCharts();
+        
+        // Setup event listeners
+        this.setupEventListeners();
+        
+        // Setup WebSocket subscriptions
+        this.setupWebSocketSubscriptions();
+        
+        // Start simulated updates for demo
+        this.startDemoUpdates();
+    }
+
+    async loadSystemMetrics() {
+        const container = document.getElementById('system-metrics');
+        if (!container) return;
+        
+        const metrics = [
+            { label: 'CPU Usage', value: '24%', trend: 'down', color: 'success' },
+            { label: 'Memory Usage', value: '1.8 GB', trend: 'stable', color: 'warning' },
+            { label: 'Disk I/O', value: '45 MB/s', trend: 'up', color: 'primary' },
+            { label: 'Network', value: '125 KB/s', trend: 'up', color: 'info' },
+            { label: 'Active Connections', value: '42', trend: 'stable', color: 'success' },
+            { label: 'Queue Length', value: '3', trend: 'down', color: 'success' }
+        ];
+        
+        container.innerHTML = metrics.map(metric => `
+            <div class="metric-item">
+                <div class="metric-label">${metric.label}</div>
+                <div class="metric-value">${metric.value}</div>
+                <div class="metric-trend ${metric.trend}">
+                    <i class="fas fa-arrow-${metric.trend === 'up' ? 'up' : metric.trend === 'down' ? 'down' : 'minus'}"></i>
+                </div>
+            </div>
+        `).join('');
+        
+        // Add CSS for metrics
+        this.addMetricStyles();
+    }
+
+    addMetricStyles() {
+        const style = document.createElement('style');
+        style.textContent = `
+            .metric-item {
+                display: flex;
+                align-items: center;
+                justify-content: space-between;
+                padding: 0.75rem;
+                border-bottom: 1px solid var(--border-color);
+            }
+            
+            .metric-item:last-child {
+                border-bottom: none;
+            }
+            
+            .metric-label {
+                font-size: 0.875rem;
+                color: var(--text-secondary);
+            }
+            
+            .metric-value {
+                font-weight: 600;
+                color: var(--text-primary);
+            }
+            
+            .metric-trend {
+                width: 24px;
+                height: 24px;
+                border-radius: 50%;
+                display: flex;
+                align-items: center;
+                justify-content: center;
+                font-size: 0.75rem;
+            }
+            
+            .metric-trend.up {
+                background-color: rgba(239, 68, 68, 0.1);
+                color: var(--danger);
+            }
+            
+            .metric-trend.down {
+                background-color: rgba(16, 185, 129, 0.1);
+                color: var(--success);
+            }
+            
+            .metric-trend.stable {
+                background-color: rgba(100, 116, 139, 0.1);
+                color: var(--text-secondary);
+            }
+            
+            .monitoring-stream {
+                height: 300px;
+                overflow-y: auto;
+                border: 1px solid var(--border-color);
+                border-radius: var(--border-radius-sm);
+                padding: 0.5rem;
+                background-color: var(--bg-secondary);
+            }
+            
+            .stream-entry {
+                display: flex;
+                align-items: center;
+                gap: 0.75rem;
+                padding: 0.5rem;
+                border-bottom: 1px solid var(--border-color);
+                transition: background-color 0.2s ease;
+            }
+            
+            .stream-entry:last-child {
+                border-bottom: none;
+            }
+            
+            .stream-entry.highlight {
+                background-color: rgba(37, 99, 235, 0.05);
+            }
+            
+            .stream-entry-time {
+                font-size: 0.75rem;
+                color: var(--text-light);
+                min-width: 60px;
+            }
+            
+            .stream-entry-icon {
+                font-size: 0.875rem;
+                width: 24px;
+                text-align: center;
+            }
+            
+            .stream-entry-content {
+                flex: 1;
+                font-size: 0.875rem;
+            }
+            
+            .stream-entry-details {
+                font-size: 0.75rem;
+                color: var(--text-secondary);
+                margin-top: 0.125rem;
+            }
+            
+            .log-stream {
+                height: 400px;
+                overflow-y: auto;
+                border: 1px solid var(--border-color);
+                border-radius: var(--border-radius-sm);
+                padding: 0.5rem;
+                background-color: var(--bg-secondary);
+                font-family: monospace;
+                font-size: 0.75rem;
+            }
+            
+            .log-entry {
+                display: flex;
+                align-items: center;
+                gap: 0.75rem;
+                padding: 0.25rem 0.5rem;
+                border-bottom: 1px solid var(--border-color);
+            }
+            
+            .log-entry:last-child {
+                border-bottom: none;
+            }
+            
+            .log-time {
+                color: var(--text-light);
+                min-width: 80px;
+            }
+            
+            .log-level {
+                width: 24px;
+                text-align: center;
+            }
+            
+            .log-info .log-level {
+                color: var(--info);
+            }
+            
+            .log-warn .log-level {
+                color: var(--warning);
+            }
+            
+            .log-error .log-level {
+                color: var(--danger);
+            }
+            
+            .log-debug .log-level {
+                color: var(--text-light);
+            }
+            
+            .log-message {
+                flex: 1;
+                white-space: nowrap;
+                overflow: hidden;
+                text-overflow: ellipsis;
+            }
+        `;
+        document.head.appendChild(style);
+    }
+
+    async loadCharts() {
+        await this.loadResponseTimeChart();
+        await this.loadErrorRateChart();
+        await this.loadRateLimitChart();
+    }
+
+    async loadResponseTimeChart() {
+        try {
+            // Generate demo data for response time
+            const labels = Array.from({ length: 20 }, (_, i) => `${i + 1}m`);
+            const data = {
+                labels: labels,
+                datasets: [{
+                    label: 'Response Time (ms)',
+                    data: labels.map(() => Math.floor(Math.random() * 200) + 300),
+                    color: '#3b82f6',
+                    fill: true
+                }]
+            };
+            
+            window.chartManager.createLineChart('response-time-chart', data, {
+                scales: {
+                    y: {
+                        title: {
+                            display: true,
+                            text: 'Milliseconds'
+                        }
+                    }
+                }
+            });
+            
+        } catch (error) {
+            console.error('Error loading response time chart:', error);
+        }
+    }
+
+    async loadErrorRateChart() {
+        try {
+            const labels = Array.from({ length: 20 }, (_, i) => `${i + 1}m`);
+            const data = {
+                labels: labels,
+                datasets: [{
+                    label: 'Error Rate (%)',
+                    data: labels.map(() => Math.random() * 5),
+                    color: '#ef4444',
+                    fill: true
+                }]
+            };
+            
+            window.chartManager.createLineChart('error-rate-chart', data, {
+                scales: {
+                    y: {
+                        title: {
+                            display: true,
+                            text: 'Percentage'
+                        },
+                        ticks: {
+                            callback: function(value) {
+                                return value + '%';
+                            }
+                        }
+                    }
+                }
+            });
+            
+        } catch (error) {
+            console.error('Error loading error rate chart:', error);
+        }
+    }
+
+    async loadRateLimitChart() {
+        try {
+            const labels = ['Web App', 'Mobile App', 'API Integration', 'Internal Tools', 'Testing'];
+            const data = {
+                labels: labels,
+                datasets: [{
+                    label: 'Rate Limit Usage',
+                    data: [65, 45, 78, 34, 60],
+                    color: '#10b981'
+                }]
+            };
+            
+            window.chartManager.createBarChart('rate-limit-chart', data, {
+                scales: {
+                    y: {
+                        title: {
+                            display: true,
+                            text: 'Percentage'
+                        },
+                        ticks: {
+                            callback: function(value) {
+                                return value + '%';
+                            }
+                        }
+                    }
+                }
+            });
+            
+        } catch (error) {
+            console.error('Error loading rate limit chart:', error);
+        }
+    }
+
+    setupEventListeners() {
+        // Pause/resume monitoring button
+        const pauseBtn = document.getElementById('pause-monitoring');
+        if (pauseBtn) {
+            pauseBtn.addEventListener('click', () => {
+                this.togglePause();
+            });
+        }
+    }
+
+    setupWebSocketSubscriptions() {
+        if (!window.wsManager) return;
+        
+        // Subscribe to request updates
+        window.wsManager.subscribe('requests', (request) => {
+            if (!this.isPaused) {
+                this.addToRequestStream(request);
+            }
+        });
+        
+        // Subscribe to log updates
+        window.wsManager.subscribe('logs', (log) => {
+            if (!this.isPaused) {
+                this.addToLogStream(log);
+            }
+        });
+        
+        // Subscribe to metric updates
+        window.wsManager.subscribe('metrics', (metric) => {
+            if (!this.isPaused) {
+                this.updateCharts(metric);
+            }
+        });
+    }
+
+    togglePause() {
+        this.isPaused = !this.isPaused;
+        const pauseBtn = document.getElementById('pause-monitoring');
+        
+        if (pauseBtn) {
+            if (this.isPaused) {
+                pauseBtn.innerHTML = '<i class="fas fa-play"></i> Resume';
+                pauseBtn.classList.remove('btn-secondary');
+                pauseBtn.classList.add('btn-success');
+                
+                if (window.authManager) {
+                    window.authManager.showToast('Monitoring paused', 'warning');
+                }
+            } else {
+                pauseBtn.innerHTML = '<i class="fas fa-pause"></i> Pause';
+                pauseBtn.classList.remove('btn-success');
+                pauseBtn.classList.add('btn-secondary');
+                
+                if (window.authManager) {
+                    window.authManager.showToast('Monitoring resumed', 'success');
+                }
+            }
+        }
+    }
+
+    addToRequestStream(request) {
+        const streamElement = document.getElementById('request-stream');
+        if (!streamElement) return;
+        
+        // Create entry
+        const entry = document.createElement('div');
+        entry.className = 'stream-entry';
+        
+        // Format time
+        const time = new Date().toLocaleTimeString();
+        
+        // Determine icon based on status
+        let icon = 'question-circle';
+        let color = 'var(--text-secondary)';
+        
+        if (request.status === 'success') {
+            icon = 'check-circle';
+            color = 'var(--success)';
+        } else if (request.status === 'error') {
+            icon = 'exclamation-circle';
+            color = 'var(--danger)';
+        }
+        
+        entry.innerHTML = `
+            <div class="stream-entry-time">${time}</div>
+            <div class="stream-entry-icon" style="color: ${color}">
+                <i class="fas fa-${icon}"></i>
+            </div>
+            <div class="stream-entry-content">
+                <strong>${request.client_id || 'Unknown'}</strong> → 
+                ${request.provider || 'Unknown'} (${request.model || 'Unknown'})
+                <div class="stream-entry-details">
+                    ${request.tokens || 0} tokens • ${request.duration || 0}ms
+                </div>
+            </div>
+        `;
+        
+        // Add to top of stream
+        streamElement.insertBefore(entry, streamElement.firstChild);
+        
+        // Store in memory (limit to 100)
+        this.requestStream.unshift({
+            time,
+            request,
+            element: entry
+        });
+        
+        if (this.requestStream.length > 100) {
+            const oldEntry = this.requestStream.pop();
+            if (oldEntry.element.parentNode) {
+                oldEntry.element.remove();
+            }
+        }
+        
+        // Add highlight animation
+        entry.classList.add('highlight');
+        setTimeout(() => entry.classList.remove('highlight'), 1000);
+    }
+
+    addToLogStream(log) {
+        const logElement = document.getElementById('system-logs');
+        if (!logElement) return;
+        
+        // Create entry
+        const entry = document.createElement('div');
+        entry.className = `log-entry log-${log.level || 'info'}`;
+        
+        // Format time
+        const time = new Date().toLocaleTimeString();
+        
+        // Determine icon based on level
+        let icon = 'info-circle';
+        if (log.level === 'error') icon = 'exclamation-circle';
+        if (log.level === 'warn') icon = 'exclamation-triangle';
+        if (log.level === 'debug') icon = 'bug';
+        
+        entry.innerHTML = `
+            <div class="log-time">${time}</div>
+            <div class="log-level">
+                <i class="fas fa-${icon}"></i>
+            </div>
+            <div class="log-message">${log.message || ''}</div>
+        `;
+        
+        // Add to top of stream
+        logElement.insertBefore(entry, logElement.firstChild);
+        
+        // Store in memory (limit to 100)
+        this.systemLogs.unshift({
+            time,
+            log,
+            element: entry
+        });
+        
+        if (this.systemLogs.length > 100) {
+            const oldEntry = this.systemLogs.pop();
+            if (oldEntry.element.parentNode) {
+                oldEntry.element.remove();
+            }
+        }
+    }
+
+    updateCharts(metric) {
+        // Update charts with new metric data
+        if (metric.type === 'response_time' && window.chartManager.charts.has('response-time-chart')) {
+            this.updateResponseTimeChart(metric.value);
+        }
+        
+        if (metric.type === 'error_rate' && window.chartManager.charts.has('error-rate-chart')) {
+            this.updateErrorRateChart(metric.value);
+        }
+    }
+
+    updateResponseTimeChart(value) {
+        window.chartManager.addDataPoint('response-time-chart', value);
+    }
+
+    updateErrorRateChart(value) {
+        window.chartManager.addDataPoint('error-rate-chart', value);
+    }
+
+    startDemoUpdates() {
+        // Simulate incoming requests for demo purposes
+        if (window.location.hostname === 'localhost' || window.location.hostname === '127.0.0.1') {
+            setInterval(() => {
+                if (!this.isPaused && Math.random() > 0.3) { // 70% chance
+                    this.simulateRequest();
+                }
+            }, 2000);
+            
+            // Simulate logs
+            setInterval(() => {
+                if (!this.isPaused && Math.random() > 0.5) { // 50% chance
+                    this.simulateLog();
+                }
+            }, 3000);
+            
+            // Simulate metrics
+            setInterval(() => {
+                if (!this.isPaused) {
+                    this.simulateMetric();
+                }
+            }, 5000);
+        }
+    }
+
+    simulateRequest() {
+        const clients = ['client-1', 'client-2', 'client-3', 'client-4', 'client-5'];
+        const providers = ['OpenAI', 'Gemini', 'DeepSeek', 'Grok'];
+        const models = ['gpt-4', 'gpt-3.5-turbo', 'gemini-pro', 'deepseek-chat', 'grok-beta'];
+        const statuses = ['success', 'success', 'success', 'error', 'warning']; // Mostly success
+        
+        const request = {
+            client_id: clients[Math.floor(Math.random() * clients.length)],
+            provider: providers[Math.floor(Math.random() * providers.length)],
+            model: models[Math.floor(Math.random() * models.length)],
+            tokens: Math.floor(Math.random() * 2000) + 100,
+            duration: Math.floor(Math.random() * 1000) + 100,
+            status: statuses[Math.floor(Math.random() * statuses.length)],
+            timestamp: Date.now()
+        };
+        
+        this.addToRequestStream(request);
+    }
+
+    simulateLog() {
+        const levels = ['info', 'info', 'info', 'warn', 'error'];
+        const messages = [
+            'Request processed successfully',
+            'Cache hit for model gpt-4',
+            'Rate limit check passed',
+            'High latency detected for DeepSeek provider',
+            'API key validation failed',
+            'Database connection pool healthy',
+            'New client registered: client-7',
+            'Backup completed successfully',
+            'Memory usage above 80% threshold',
+            'Provider Grok is offline'
+        ];
+        
+        const log = {
+            level: levels[Math.floor(Math.random() * levels.length)],
+            message: messages[Math.floor(Math.random() * messages.length)],
+            timestamp: Date.now()
+        };
+        
+        this.addToLogStream(log);
+    }
+
+    simulateMetric() {
+        const metricTypes = ['response_time', 'error_rate'];
+        const type = metricTypes[Math.floor(Math.random() * metricTypes.length)];
+        
+        let value;
+        if (type === 'response_time') {
+            value = Math.floor(Math.random() * 200) + 300; // 300-500ms
+        } else {
+            value = Math.random() * 5; // 0-5%
+        }
+        
+        this.updateCharts({ type, value });
+    }
+
+    clearStreams() {
+        const streamElement = document.getElementById('request-stream');
+        const logElement = document.getElementById('system-logs');
+        
+        if (streamElement) {
+            streamElement.innerHTML = '';
+            this.requestStream = [];
+        }
+        
+        if (logElement) {
+            logElement.innerHTML = '';
+            this.systemLogs = [];
+        }
+    }
+
+    refresh() {
+        this.loadSystemMetrics();
+        this.loadCharts();
+        this.clearStreams();
+        
+        if (window.authManager) {
+            window.authManager.showToast('Monitoring refreshed', 'success');
+        }
+    }
+}
+
+// Initialize monitoring page when needed
+window.initMonitoring = async () => {
+    window.monitoringPage = new MonitoringPage();
+};
+
+// Export for use in other modules
+if (typeof module !== 'undefined' && module.exports) {
+    module.exports = MonitoringPage;
+}

static/js/pages/overview.js

@@ -0,0 +1,513 @@
+// Overview Page Module
+
+class OverviewPage {
+    constructor() {
+        this.stats = null;
+        this.charts = {};
+        this.init();
+    }
+
+    async init() {
+        // Load data
+        await this.loadStats();
+        await this.loadCharts();
+        await this.loadRecentRequests();
+        
+        // Setup event listeners
+        this.setupEventListeners();
+        
+        // Subscribe to WebSocket updates
+        this.setupWebSocketSubscriptions();
+    }
+
+    async loadStats() {
+        try {
+            // In a real app, this would fetch from /api/usage/summary
+            // For now, use mock data
+            this.stats = {
+                totalRequests: 12458,
+                totalTokens: 1254300,
+                totalCost: 125.43,
+                activeClients: 8,
+                errorRate: 2.3,
+                avgResponseTime: 450,
+                todayRequests: 342,
+                todayCost: 12.45
+            };
+            
+            this.renderStats();
+            
+        } catch (error) {
+            console.error('Error loading stats:', error);
+            this.showError('Failed to load statistics');
+        }
+    }
+
+    renderStats() {
+        const container = document.getElementById('overview-stats');
+        if (!container) return;
+        
+        container.innerHTML = `
+            <div class="stat-card">
+                <div class="stat-icon primary">
+                    <i class="fas fa-exchange-alt"></i>
+                </div>
+                <div class="stat-content">
+                    <div class="stat-value">${this.stats.totalRequests.toLocaleString()}</div>
+                    <div class="stat-label">Total Requests</div>
+                    <div class="stat-change positive">
+                        <i class="fas fa-arrow-up"></i>
+                        ${this.stats.todayRequests} today
+                    </div>
+                </div>
+            </div>
+            
+            <div class="stat-card">
+                <div class="stat-icon success">
+                    <i class="fas fa-coins"></i>
+                </div>
+                <div class="stat-content">
+                    <div class="stat-value">${this.stats.totalTokens.toLocaleString()}</div>
+                    <div class="stat-label">Total Tokens</div>
+                    <div class="stat-change positive">
+                        <i class="fas fa-arrow-up"></i>
+                        12% from yesterday
+                    </div>
+                </div>
+            </div>
+            
+            <div class="stat-card">
+                <div class="stat-icon warning">
+                    <i class="fas fa-dollar-sign"></i>
+                </div>
+                <div class="stat-content">
+                    <div class="stat-value">$${this.stats.totalCost.toFixed(2)}</div>
+                    <div class="stat-label">Total Cost</div>
+                    <div class="stat-change positive">
+                        <i class="fas fa-arrow-up"></i>
+                        $${this.stats.todayCost.toFixed(2)} today
+                    </div>
+                </div>
+            </div>
+            
+            <div class="stat-card">
+                <div class="stat-icon danger">
+                    <i class="fas fa-users"></i>
+                </div>
+                <div class="stat-content">
+                    <div class="stat-value">${this.stats.activeClients}</div>
+                    <div class="stat-label">Active Clients</div>
+                    <div class="stat-change positive">
+                        <i class="fas fa-arrow-up"></i>
+                        2 new this week
+                    </div>
+                </div>
+            </div>
+            
+            <div class="stat-card">
+                <div class="stat-icon primary">
+                    <i class="fas fa-exclamation-triangle"></i>
+                </div>
+                <div class="stat-content">
+                    <div class="stat-value">${this.stats.errorRate}%</div>
+                    <div class="stat-label">Error Rate</div>
+                    <div class="stat-change negative">
+                        <i class="fas fa-arrow-down"></i>
+                        0.5% improvement
+                    </div>
+                </div>
+            </div>
+            
+            <div class="stat-card">
+                <div class="stat-icon success">
+                    <i class="fas fa-tachometer-alt"></i>
+                </div>
+                <div class="stat-content">
+                    <div class="stat-value">${this.stats.avgResponseTime}ms</div>
+                    <div class="stat-label">Avg Response Time</div>
+                    <div class="stat-change positive">
+                        <i class="fas fa-arrow-down"></i>
+                        50ms faster
+                    </div>
+                </div>
+            </div>
+        `;
+    }
+
+    async loadCharts() {
+        await this.loadRequestsChart();
+        await this.loadProvidersChart();
+        await this.loadSystemHealth();
+    }
+
+    async loadRequestsChart() {
+        try {
+            // Generate demo data for requests chart
+            const data = window.chartManager.generateDemoTimeSeries(24, 1);
+            data.datasets[0].label = 'Requests per hour';
+            data.datasets[0].fill = true;
+            
+            // Create chart
+            this.charts.requests = window.chartManager.createLineChart('requests-chart', data, {
+                plugins: {
+                    tooltip: {
+                        callbacks: {
+                            label: function(context) {
+                                return `Requests: ${context.parsed.y}`;
+                            }
+                        }
+                    }
+                }
+            });
+            
+        } catch (error) {
+            console.error('Error loading requests chart:', error);
+        }
+    }
+
+    async loadProvidersChart() {
+        try {
+            const data = {
+                labels: ['OpenAI', 'Gemini', 'DeepSeek', 'Grok'],
+                data: [45, 25, 20, 10],
+                colors: ['#3b82f6', '#10b981', '#f59e0b', '#8b5cf6']
+            };
+            
+            this.charts.providers = window.chartManager.createDoughnutChart('providers-chart', data, {
+                plugins: {
+                    tooltip: {
+                        callbacks: {
+                            label: function(context) {
+                                const label = context.label || '';
+                                const value = context.raw || 0;
+                                return `${label}: ${value}% of requests`;
+                            }
+                        }
+                    }
+                }
+            });
+            
+        } catch (error) {
+            console.error('Error loading providers chart:', error);
+        }
+    }
+
+    async loadSystemHealth() {
+        const container = document.getElementById('system-health');
+        if (!container) return;
+        
+        const healthData = [
+            { label: 'API Server', status: 'online', value: 100 },
+            { label: 'Database', status: 'online', value: 95 },
+            { label: 'OpenAI', status: 'online', value: 100 },
+            { label: 'Gemini', status: 'online', value: 100 },
+            { label: 'DeepSeek', status: 'warning', value: 85 },
+            { label: 'Grok', status: 'offline', value: 0 }
+        ];
+        
+        container.innerHTML = healthData.map(item => `
+            <div class="health-item">
+                <div class="health-label">
+                    <span class="health-status status-badge ${item.status}">
+                        <i class="fas fa-circle"></i>
+                        ${item.status}
+                    </span>
+                    <span class="health-name">${item.label}</span>
+                </div>
+                <div class="health-progress">
+                    <div class="progress-bar">
+                        <div class="progress-fill ${item.status}" style="width: ${item.value}%"></div>
+                    </div>
+                    <span class="health-value">${item.value}%</span>
+                </div>
+            </div>
+        `).join('');
+        
+        // Add CSS for progress bars
+        this.addHealthStyles();
+    }
+
+    addHealthStyles() {
+        const style = document.createElement('style');
+        style.textContent = `
+            .health-item {
+                margin-bottom: 1rem;
+            }
+            
+            .health-label {
+                display: flex;
+                align-items: center;
+                gap: 0.5rem;
+                margin-bottom: 0.25rem;
+            }
+            
+            .health-name {
+                font-size: 0.875rem;
+                color: var(--text-primary);
+            }
+            
+            .health-progress {
+                display: flex;
+                align-items: center;
+                gap: 0.5rem;
+            }
+            
+            .progress-bar {
+                flex: 1;
+                height: 6px;
+                background-color: var(--bg-secondary);
+                border-radius: 3px;
+                overflow: hidden;
+            }
+            
+            .progress-fill {
+                height: 100%;
+                border-radius: 3px;
+                transition: width 0.3s ease;
+            }
+            
+            .progress-fill.online {
+                background-color: var(--success);
+            }
+            
+            .progress-fill.warning {
+                background-color: var(--warning);
+            }
+            
+            .progress-fill.offline {
+                background-color: var(--danger);
+            }
+            
+            .health-value {
+                font-size: 0.75rem;
+                color: var(--text-secondary);
+                min-width: 40px;
+                text-align: right;
+            }
+        `;
+        document.head.appendChild(style);
+    }
+
+    async loadRecentRequests() {
+        try {
+            // In a real app, this would fetch from /api/requests/recent
+            // For now, use mock data
+            const requests = [
+                { time: '14:32:15', client: 'client-1', provider: 'OpenAI', model: 'gpt-4', tokens: 1250, status: 'success' },
+                { time: '14:30:45', client: 'client-2', provider: 'Gemini', model: 'gemini-pro', tokens: 890, status: 'success' },
+                { time: '14:28:12', client: 'client-3', provider: 'DeepSeek', model: 'deepseek-chat', tokens: 1560, status: 'error' },
+                { time: '14:25:33', client: 'client-1', provider: 'OpenAI', model: 'gpt-3.5-turbo', tokens: 540, status: 'success' },
+                { time: '14:22:18', client: 'client-4', provider: 'Grok', model: 'grok-beta', tokens: 720, status: 'success' },
+                { time: '14:20:05', client: 'client-2', provider: 'Gemini', model: 'gemini-pro-vision', tokens: 1120, status: 'success' },
+                { time: '14:18:47', client: 'client-5', provider: 'OpenAI', model: 'gpt-4', tokens: 980, status: 'warning' },
+                { time: '14:15:22', client: 'client-3', provider: 'DeepSeek', model: 'deepseek-coder', tokens: 1340, status: 'success' },
+                { time: '14:12:10', client: 'client-1', provider: 'OpenAI', model: 'gpt-3.5-turbo', tokens: 610, status: 'success' },
+                { time: '14:10:05', client: 'client-6', provider: 'Gemini', model: 'gemini-pro', tokens: 830, status: 'success' }
+            ];
+            
+            this.renderRecentRequests(requests);
+            
+        } catch (error) {
+            console.error('Error loading recent requests:', error);
+        }
+    }
+
+    renderRecentRequests(requests) {
+        const tableBody = document.querySelector('#recent-requests tbody');
+        if (!tableBody) return;
+        
+        tableBody.innerHTML = requests.map(request => {
+            const statusClass = request.status === 'success' ? 'success' : 
+                              request.status === 'error' ? 'danger' : 'warning';
+            const statusIcon = request.status === 'success' ? 'check-circle' :
+                              request.status === 'error' ? 'exclamation-circle' : 'exclamation-triangle';
+            
+            return `
+                <tr>
+                    <td>${request.time}</td>
+                    <td>${request.client}</td>
+                    <td>${request.provider}</td>
+                    <td>${request.model}</td>
+                    <td>${request.tokens.toLocaleString()}</td>
+                    <td>
+                        <span class="status-badge ${statusClass}">
+                            <i class="fas fa-${statusIcon}"></i>
+                            ${request.status}
+                        </span>
+                    </td>
+                </tr>
+            `;
+        }).join('');
+    }
+
+    setupEventListeners() {
+        // Period buttons for requests chart
+        const periodButtons = document.querySelectorAll('.chart-control-btn[data-period]');
+        periodButtons.forEach(button => {
+            button.addEventListener('click', () => {
+                // Update active state
+                periodButtons.forEach(btn => btn.classList.remove('active'));
+                button.classList.add('active');
+                
+                // Update chart based on period
+                this.updateRequestsChart(button.dataset.period);
+            });
+        });
+        
+        // Refresh button for recent requests
+        const refreshBtn = document.querySelector('#recent-requests .card-action-btn');
+        if (refreshBtn) {
+            refreshBtn.addEventListener('click', () => {
+                this.loadRecentRequests();
+                if (window.authManager) {
+                    window.authManager.showToast('Recent requests refreshed', 'success');
+                }
+            });
+        }
+    }
+
+    setupWebSocketSubscriptions() {
+        if (!window.wsManager) return;
+        
+        // Subscribe to request updates
+        window.wsManager.subscribe('requests', (request) => {
+            this.handleNewRequest(request);
+        });
+        
+        // Subscribe to metric updates
+        window.wsManager.subscribe('metrics', (metric) => {
+            this.handleNewMetric(metric);
+        });
+    }
+
+    handleNewRequest(request) {
+        // Update total requests counter
+        if (this.stats) {
+            this.stats.totalRequests++;
+            this.stats.todayRequests++;
+            
+            // Update tokens if available
+            if (request.tokens) {
+                this.stats.totalTokens += request.tokens;
+            }
+            
+            // Re-render stats
+            this.renderStats();
+        }
+        
+        // Add to recent requests table
+        this.addToRecentRequests(request);
+    }
+
+    addToRecentRequests(request) {
+        const tableBody = document.querySelector('#recent-requests tbody');
+        if (!tableBody) return;
+        
+        const time = new Date(request.timestamp || Date.now()).toLocaleTimeString();
+        const statusClass = request.status === 'success' ? 'success' : 
+                          request.status === 'error' ? 'danger' : 'warning';
+        const statusIcon = request.status === 'success' ? 'check-circle' :
+                          request.status === 'error' ? 'exclamation-circle' : 'exclamation-triangle';
+        
+        const row = document.createElement('tr');
+        row.innerHTML = `
+            <td>${time}</td>
+            <td>${request.client_id || 'Unknown'}</td>
+            <td>${request.provider || 'Unknown'}</td>
+            <td>${request.model || 'Unknown'}</td>
+            <td>${request.tokens || 0}</td>
+            <td>
+                <span class="status-badge ${statusClass}">
+                    <i class="fas fa-${statusIcon}"></i>
+                    ${request.status || 'unknown'}
+                </span>
+            </td>
+        `;
+        
+        // Add to top of table
+        tableBody.insertBefore(row, tableBody.firstChild);
+        
+        // Limit to 50 rows
+        const rows = tableBody.querySelectorAll('tr');
+        if (rows.length > 50) {
+            tableBody.removeChild(rows[rows.length - 1]);
+        }
+    }
+
+    handleNewMetric(metric) {
+        // Update charts with new metric data
+        if (metric.type === 'requests' && this.charts.requests) {
+            this.updateRequestsChartData(metric);
+        }
+        
+        // Update system health if needed
+        if (metric.type === 'system_health') {
+            this.updateSystemHealth(metric);
+        }
+    }
+
+    updateRequestsChart(period) {
+        // In a real app, this would fetch new data based on period
+        // For now, just update with demo data
+        let hours = 24;
+        if (period === '7d') hours = 24 * 7;
+        if (period === '30d') hours = 24 * 30;
+        
+        const data = window.chartManager.generateDemoTimeSeries(hours, 1);
+        data.datasets[0].label = 'Requests';
+        data.datasets[0].fill = true;
+        
+        window.chartManager.updateChartData('requests-chart', data);
+    }
+
+    updateRequestsChartData(metric) {
+        // Add new data point to the chart
+        if (this.charts.requests && metric.value !== undefined) {
+            window.chartManager.addDataPoint('requests-chart', metric.value);
+        }
+    }
+
+    updateSystemHealth(metric) {
+        // Update system health indicators
+        const container = document.getElementById('system-health');
+        if (!container || !metric.data) return;
+        
+        // This would update specific health indicators based on metric data
+        // Implementation depends on metric structure
+    }
+
+    showError(message) {
+        const container = document.getElementById('overview-stats');
+        if (container) {
+            container.innerHTML = `
+                <div class="error-message" style="grid-column: 1 / -1;">
+                    <i class="fas fa-exclamation-circle"></i>
+                    <span>${message}</span>
+                </div>
+            `;
+        }
+    }
+
+    refresh() {
+        this.loadStats();
+        this.loadRecentRequests();
+        
+        // Refresh charts
+        if (this.charts.requests) {
+            this.charts.requests.update();
+        }
+        if (this.charts.providers) {
+            this.charts.providers.update();
+        }
+    }
+}
+
+// Initialize overview page when needed
+window.initOverview = async () => {
+    window.overviewPage = new OverviewPage();
+};
+
+// Export for use in other modules
+if (typeof module !== 'undefined' && module.exports) {
+    module.exports = OverviewPage;
+}

static/js/pages/providers.js

@@ -0,0 +1,650 @@
+// Providers Page Module
+
+class ProvidersPage {
+    constructor() {
+        this.providers = [];
+        this.init();
+    }
+
+    async init() {
+        // Load data
+        await this.loadProviderStats();
+        await this.loadProvidersList();
+        await this.loadModelsList();
+        await this.loadConnectionTests();
+        
+        // Setup event listeners
+        this.setupEventListeners();
+    }
+
+    async loadProviderStats() {
+        const container = document.getElementById('provider-stats');
+        if (!container) return;
+        
+        container.innerHTML = `
+            <div class="stat-card">
+                <div class="stat-icon primary">
+                    <i class="fas fa-server"></i>
+                </div>
+                <div class="stat-content">
+                    <div class="stat-value">4</div>
+                    <div class="stat-label">Total Providers</div>
+                    <div class="stat-change">
+                        <i class="fas fa-check-circle"></i>
+                        3 active
+                    </div>
+                </div>
+            </div>
+            
+            <div class="stat-card">
+                <div class="stat-icon success">
+                    <i class="fas fa-plug"></i>
+                </div>
+                <div class="stat-content">
+                    <div class="stat-value">3</div>
+                    <div class="stat-label">Connected</div>
+                    <div class="stat-change positive">
+                        <i class="fas fa-arrow-up"></i>
+                        All systems operational
+                    </div>
+                </div>
+            </div>
+            
+            <div class="stat-card">
+                <div class="stat-icon warning">
+                    <i class="fas fa-exclamation-triangle"></i>
+                </div>
+                <div class="stat-content">
+                    <div class="stat-value">1</div>
+                    <div class="stat-label">Issues</div>
+                    <div class="stat-change">
+                        <i class="fas fa-info-circle"></i>
+                        DeepSeek: 85% health
+                    </div>
+                </div>
+            </div>
+            
+            <div class="stat-card">
+                <div class="stat-icon danger">
+                    <i class="fas fa-times-circle"></i>
+                </div>
+                <div class="stat-content">
+                    <div class="stat-value">1</div>
+                    <div class="stat-label">Offline</div>
+                    <div class="stat-change">
+                        <i class="fas fa-redo"></i>
+                        Grok: Connection failed
+                    </div>
+                </div>
+            </div>
+        `;
+    }
+
+    async loadProvidersList() {
+        const container = document.getElementById('providers-list');
+        if (!container) return;
+        
+        this.providers = [
+            { name: 'OpenAI', enabled: true, status: 'online', apiKey: 'sk-*****123', models: ['gpt-4', 'gpt-3.5-turbo'], lastUsed: '2024-01-15 14:32:15' },
+            { name: 'Gemini', enabled: true, status: 'online', apiKey: 'AIza*****456', models: ['gemini-pro', 'gemini-pro-vision'], lastUsed: '2024-01-15 14:30:45' },
+            { name: 'DeepSeek', enabled: true, status: 'warning', apiKey: 'sk-*****789', models: ['deepseek-chat', 'deepseek-coder'], lastUsed: '2024-01-15 14:28:12' },
+            { name: 'Grok', enabled: false, status: 'offline', apiKey: 'gk-*****012', models: ['grok-beta'], lastUsed: '2024-01-12 10:15:22' }
+        ];
+        
+        container.innerHTML = this.providers.map(provider => {
+            const statusClass = provider.status === 'online' ? 'success' : 
+                              provider.status === 'warning' ? 'warning' : 'danger';
+            const statusIcon = provider.status === 'online' ? 'check-circle' :
+                              provider.status === 'warning' ? 'exclamation-triangle' : 'times-circle';
+            
+            return `
+                <div class="provider-card">
+                    <div class="provider-header">
+                        <div class="provider-info">
+                            <h4 class="provider-name">${provider.name}</h4>
+                            <span class="status-badge ${statusClass}">
+                                <i class="fas fa-${statusIcon}"></i>
+                                ${provider.status}
+                            </span>
+                        </div>
+                        <div class="provider-actions">
+                            <label class="toggle-switch">
+                                <input type="checkbox" ${provider.enabled ? 'checked' : ''} data-provider="${provider.name}">
+                                <span class="toggle-slider"></span>
+                            </label>
+                            <button class="btn-action" title="Configure" data-action="configure" data-provider="${provider.name}">
+                                <i class="fas fa-cog"></i>
+                            </button>
+                            <button class="btn-action" title="Test Connection" data-action="test" data-provider="${provider.name}">
+                                <i class="fas fa-play"></i>
+                            </button>
+                        </div>
+                    </div>
+                    <div class="provider-details">
+                        <div class="detail-item">
+                            <span class="detail-label">API Key:</span>
+                            <code class="detail-value">${provider.apiKey}</code>
+                            <button class="btn-copy" data-text="${provider.apiKey}" title="Copy">
+                                <i class="fas fa-copy"></i>
+                            </button>
+                        </div>
+                        <div class="detail-item">
+                            <span class="detail-label">Models:</span>
+                            <span class="detail-value">${provider.models.join(', ')}</span>
+                        </div>
+                        <div class="detail-item">
+                            <span class="detail-label">Last Used:</span>
+                            <span class="detail-value">${provider.lastUsed}</span>
+                        </div>
+                    </div>
+                </div>
+            `;
+        }).join('');
+        
+        // Add CSS for provider cards
+        this.addProviderStyles();
+    }
+
+    addProviderStyles() {
+        const style = document.createElement('style');
+        style.textContent = `
+            .provider-card {
+                background-color: var(--bg-card);
+                border: 1px solid var(--border-color);
+                border-radius: var(--border-radius);
+                padding: 1rem;
+                margin-bottom: 1rem;
+            }
+            
+            .provider-header {
+                display: flex;
+                justify-content: space-between;
+                align-items: center;
+                margin-bottom: 1rem;
+            }
+            
+            .provider-info {
+                display: flex;
+                align-items: center;
+                gap: 0.5rem;
+            }
+            
+            .provider-name {
+                font-size: 1rem;
+                font-weight: 600;
+                color: var(--text-primary);
+                margin: 0;
+            }
+            
+            .provider-actions {
+                display: flex;
+                align-items: center;
+                gap: 0.5rem;
+            }
+            
+            .toggle-switch {
+                position: relative;
+                display: inline-block;
+                width: 50px;
+                height: 24px;
+            }
+            
+            .toggle-switch input {
+                opacity: 0;
+                width: 0;
+                height: 0;
+            }
+            
+            .toggle-slider {
+                position: absolute;
+                cursor: pointer;
+                top: 0;
+                left: 0;
+                right: 0;
+                bottom: 0;
+                background-color: var(--text-light);
+                transition: .4s;
+                border-radius: 24px;
+            }
+            
+            .toggle-slider:before {
+                position: absolute;
+                content: "";
+                height: 16px;
+                width: 16px;
+                left: 4px;
+                bottom: 4px;
+                background-color: white;
+                transition: .4s;
+                border-radius: 50%;
+            }
+            
+            input:checked + .toggle-slider {
+                background-color: var(--success);
+            }
+            
+            input:checked + .toggle-slider:before {
+                transform: translateX(26px);
+            }
+            
+            .provider-details {
+                display: grid;
+                grid-template-columns: repeat(auto-fit, minmax(200px, 1fr));
+                gap: 0.75rem;
+                font-size: 0.875rem;
+            }
+            
+            .detail-item {
+                display: flex;
+                align-items: center;
+                gap: 0.5rem;
+            }
+            
+            .detail-label {
+                color: var(--text-secondary);
+                font-weight: 500;
+                min-width: 70px;
+            }
+            
+            .detail-value {
+                color: var(--text-primary);
+                flex: 1;
+            }
+            
+            .btn-copy {
+                background: none;
+                border: none;
+                color: var(--text-secondary);
+                cursor: pointer;
+                font-size: 0.75rem;
+                padding: 0.25rem;
+                transition: color 0.2s ease;
+            }
+            
+            .btn-copy:hover {
+                color: var(--primary);
+            }
+        `;
+        document.head.appendChild(style);
+    }
+
+    async loadModelsList() {
+        const container = document.getElementById('models-list');
+        if (!container) return;
+        
+        const models = [
+            { provider: 'OpenAI', name: 'gpt-4', enabled: true, context: 8192, maxTokens: 4096 },
+            { provider: 'OpenAI', name: 'gpt-3.5-turbo', enabled: true, context: 16384, maxTokens: 4096 },
+            { provider: 'Gemini', name: 'gemini-pro', enabled: true, context: 32768, maxTokens: 8192 },
+            { provider: 'Gemini', name: 'gemini-pro-vision', enabled: true, context: 32768, maxTokens: 4096 },
+            { provider: 'DeepSeek', name: 'deepseek-chat', enabled: true, context: 16384, maxTokens: 4096 },
+            { provider: 'DeepSeek', name: 'deepseek-coder', enabled: true, context: 16384, maxTokens: 4096 },
+            { provider: 'Grok', name: 'grok-beta', enabled: false, context: 8192, maxTokens: 2048 }
+        ];
+        
+        container.innerHTML = models.map(model => `
+            <div class="model-item">
+                <div class="model-header">
+                    <span class="model-name">${model.name}</span>
+                    <span class="model-provider">${model.provider}</span>
+                </div>
+                <div class="model-details">
+                    <span class="model-detail">
+                        <i class="fas fa-microchip"></i>
+                        Context: ${model.context.toLocaleString()} tokens
+                    </span>
+                    <span class="model-detail">
+                        <i class="fas fa-ruler"></i>
+                        Max: ${model.maxTokens.toLocaleString()} tokens
+                    </span>
+                    <span class="model-status ${model.enabled ? 'enabled' : 'disabled'}">
+                        <i class="fas fa-${model.enabled ? 'check' : 'times'}"></i>
+                        ${model.enabled ? 'Enabled' : 'Disabled'}
+                    </span>
+                </div>
+            </div>
+        `).join('');
+        
+        // Add CSS for model items
+        this.addModelStyles();
+    }
+
+    addModelStyles() {
+        const style = document.createElement('style');
+        style.textContent = `
+            .model-item {
+                background-color: var(--bg-secondary);
+                border-radius: var(--border-radius-sm);
+                padding: 0.75rem;
+                margin-bottom: 0.5rem;
+            }
+            
+            .model-header {
+                display: flex;
+                justify-content: space-between;
+                align-items: center;
+                margin-bottom: 0.5rem;
+            }
+            
+            .model-name {
+                font-weight: 600;
+                color: var(--text-primary);
+            }
+            
+            .model-provider {
+                font-size: 0.75rem;
+                color: var(--text-secondary);
+                background-color: var(--bg-primary);
+                padding: 0.25rem 0.5rem;
+                border-radius: 12px;
+            }
+            
+            .model-details {
+                display: flex;
+                flex-wrap: wrap;
+                gap: 1rem;
+                font-size: 0.75rem;
+            }
+            
+            .model-detail {
+                display: flex;
+                align-items: center;
+                gap: 0.25rem;
+                color: var(--text-secondary);
+            }
+            
+            .model-detail i {
+                font-size: 0.625rem;
+            }
+            
+            .model-status {
+                font-size: 0.75rem;
+                padding: 0.125rem 0.5rem;
+                border-radius: 12px;
+            }
+            
+            .model-status.enabled {
+                background-color: rgba(16, 185, 129, 0.1);
+                color: var(--success);
+            }
+            
+            .model-status.disabled {
+                background-color: rgba(239, 68, 68, 0.1);
+                color: var(--danger);
+            }
+        `;
+        document.head.appendChild(style);
+    }
+
+    async loadConnectionTests() {
+        const container = document.getElementById('connection-tests');
+        if (!container) return;
+        
+        const tests = [
+            { provider: 'OpenAI', status: 'success', latency: 245, timestamp: '2024-01-15 14:35:00' },
+            { provider: 'Gemini', status: 'success', latency: 189, timestamp: '2024-01-15 14:34:30' },
+            { provider: 'DeepSeek', status: 'warning', latency: 520, timestamp: '2024-01-15 14:34:00' },
+            { provider: 'Grok', status: 'error', latency: null, timestamp: '2024-01-15 14:33:30' }
+        ];
+        
+        container.innerHTML = tests.map(test => {
+            const statusClass = test.status === 'success' ? 'success' : 
+                              test.status === 'warning' ? 'warning' : 'danger';
+            const statusIcon = test.status === 'success' ? 'check-circle' :
+                              test.status === 'warning' ? 'exclamation-triangle' : 'times-circle';
+            
+            return `
+                <div class="test-result">
+                    <div class="test-provider">${test.provider}</div>
+                    <div class="test-status">
+                        <span class="status-badge ${statusClass}">
+                            <i class="fas fa-${statusIcon}"></i>
+                            ${test.status}
+                        </span>
+                    </div>
+                    <div class="test-latency">${test.latency ? `${test.latency}ms` : 'N/A'}</div>
+                    <div class="test-time">${test.timestamp}</div>
+                </div>
+            `;
+        }).join('');
+        
+        // Add CSS for test results
+        this.addTestStyles();
+    }
+
+    addTestStyles() {
+        const style = document.createElement('style');
+        style.textContent = `
+            .test-result {
+                display: grid;
+                grid-template-columns: 1fr 1fr 1fr 2fr;
+                gap: 1rem;
+                align-items: center;
+                padding: 0.75rem;
+                border-bottom: 1px solid var(--border-color);
+            }
+            
+            .test-result:last-child {
+                border-bottom: none;
+            }
+            
+            .test-provider {
+                font-weight: 500;
+                color: var(--text-primary);
+            }
+            
+            .test-latency {
+                color: var(--text-secondary);
+                font-family: monospace;
+            }
+            
+            .test-time {
+                color: var(--text-light);
+                font-size: 0.75rem;
+            }
+        `;
+        document.head.appendChild(style);
+    }
+
+    setupEventListeners() {
+        // Test all providers button
+        const testAllBtn = document.getElementById('test-all-providers');
+        if (testAllBtn) {
+            testAllBtn.addEventListener('click', () => {
+                this.testAllProviders();
+            });
+        }
+        
+        // Toggle switches
+        document.addEventListener('change', (e) => {
+            if (e.target.matches('.toggle-switch input')) {
+                const provider = e.target.dataset.provider;
+                const enabled = e.target.checked;
+                this.toggleProvider(provider, enabled);
+            }
+        });
+        
+        // Action buttons
+        document.addEventListener('click', (e) => {
+            if (e.target.closest('.btn-action')) {
+                const button = e.target.closest('.btn-action');
+                const action = button.dataset.action;
+                const provider = button.dataset.provider;
+                
+                switch (action) {
+                    case 'configure':
+                        this.configureProvider(provider);
+                        break;
+                    case 'test':
+                        this.testProvider(provider);
+                        break;
+                }
+            }
+            
+            // Copy buttons
+            if (e.target.closest('.btn-copy')) {
+                const button = e.target.closest('.btn-copy');
+                const text = button.dataset.text;
+                this.copyToClipboard(text);
+                
+                if (window.authManager) {
+                    window.authManager.showToast('Copied to clipboard', 'success');
+                }
+            }
+        });
+    }
+
+    toggleProvider(providerName, enabled) {
+        const provider = this.providers.find(p => p.name === providerName);
+        if (!provider) return;
+        
+        // In a real app, this would update the provider via API
+        provider.enabled = enabled;
+        provider.status = enabled ? 'online' : 'offline';
+        
+        if (window.authManager) {
+            window.authManager.showToast(
+                `${providerName} ${enabled ? 'enabled' : 'disabled'}`,
+                enabled ? 'success' : 'warning'
+            );
+        }
+        
+        // Refresh providers list
+        this.loadProvidersList();
+    }
+
+    configureProvider(providerName) {
+        const provider = this.providers.find(p => p.name === providerName);
+        if (!provider) return;
+        
+        // Show configuration modal
+        const modal = document.createElement('div');
+        modal.className = 'modal active';
+        modal.innerHTML = `
+            <div class="modal-content">
+                <div class="modal-header">
+                    <h3 class="modal-title">Configure ${providerName}</h3>
+                    <button class="modal-close">
+                        <i class="fas fa-times"></i>
+                    </button>
+                </div>
+                <div class="modal-body">
+                    <form id="configure-provider-form">
+                        <div class="form-control">
+                            <label for="api-key">API Key</label>
+                            <input type="password" id="api-key" value="${provider.apiKey}" placeholder="Enter API key" required>
+                        </div>
+                        <div class="form-control">
+                            <label for="base-url">Base URL (Optional)</label>
+                            <input type="text" id="base-url" placeholder="https://api.openai.com/v1">
+                        </div>
+                        <div class="form-control">
+                            <label for="timeout">Timeout (seconds)</label>
+                            <input type="number" id="timeout" value="30" min="1" max="300">
+                        </div>
+                        <div class="form-control">
+                            <label for="retry-count">Retry Count</label>
+                            <input type="number" id="retry-count" value="3" min="0" max="10">
+                        </div>
+                    </form>
+                </div>
+                <div class="modal-footer">
+                    <button class="btn btn-secondary close-modal">Cancel</button>
+                    <button class="btn btn-primary save-config">Save Configuration</button>
+                </div>
+            </div>
+        `;
+        
+        document.body.appendChild(modal);
+        
+        // Setup event listeners
+        const closeBtn = modal.querySelector('.modal-close');
+        const closeModalBtn = modal.querySelector('.close-modal');
+        const saveBtn = modal.querySelector('.save-config');
+        
+        const closeModal = () => {
+            modal.classList.remove('active');
+            setTimeout(() => modal.remove(), 300);
+        };
+        
+        closeBtn.addEventListener('click', closeModal);
+        closeModalBtn.addEventListener('click', closeModal);
+        
+        saveBtn.addEventListener('click', () => {
+            // In a real app, this would save provider configuration
+            if (window.authManager) {
+                window.authManager.showToast(`${providerName} configuration saved`, 'success');
+            }
+            closeModal();
+        });
+        
+        // Close on background click
+        modal.addEventListener('click', (e) => {
+            if (e.target === modal) {
+                closeModal();
+            }
+        });
+    }
+
+    testProvider(providerName) {
+        const provider = this.providers.find(p => p.name === providerName);
+        if (!provider) return;
+        
+        // Show testing in progress
+        if (window.authManager) {
+            window.authManager.showToast(`Testing ${providerName} connection...`, 'info');
+        }
+        
+        // Simulate API test
+        setTimeout(() => {
+            // In a real app, this would test the provider connection via API
+            const success = Math.random() > 0.3; // 70% success rate for demo
+            
+            if (window.authManager) {
+                window.authManager.showToast(
+                    `${providerName} connection ${success ? 'successful' : 'failed'}`,
+                    success ? 'success' : 'error'
+                );
+            }
+            
+            // Refresh connection tests
+            this.loadConnectionTests();
+        }, 1500);
+    }
+
+    testAllProviders() {
+        if (window.authManager) {
+            window.authManager.showToast('Testing all providers...', 'info');
+        }
+        
+        // Test each provider sequentially
+        this.providers.forEach((provider, index) => {
+            setTimeout(() => {
+                this.testProvider(provider.name);
+            }, index * 2000); // Stagger tests
+        });
+    }
+
+    copyToClipboard(text) {
+        navigator.clipboard.writeText(text).catch(err => {
+            console.error('Failed to copy:', err);
+        });
+    }
+
+    refresh() {
+        this.loadProviderStats();
+        this.loadProvidersList();
+        this.loadModelsList();
+        this.loadConnectionTests();
+    }
+}
+
+// Initialize providers page when needed
+window.initProviders = async () => {
+    window.providersPage = new ProvidersPage();
+};
+
+// Export for use in other modules
+if (typeof module !== 'undefined' && module.exports) {
+    module.exports = ProvidersPage;
+}

static/js/pages/settings.js

@@ -0,0 +1,318 @@
+// Settings Page Module
+
+class SettingsPage {
+    constructor() {
+        this.settings = {};
+        this.init();
+    }
+
+    async init() {
+        // Load settings
+        await this.loadSettings();
+        await this.loadSystemInfo();
+        
+        // Setup event listeners
+        this.setupEventListeners();
+    }
+
+    async loadSettings() {
+        try {
+            // In a real app, this would fetch from /api/settings
+            this.settings = {
+                serverPort: 8080,
+                logLevel: 'info',
+                dbPath: './data/llm-proxy.db',
+                backupInterval: 24,
+                sessionTimeout: 30,
+                enableRateLimiting: true,
+                enableCostTracking: true,
+                enableMetrics: true,
+                enableWebSocket: true
+            };
+            
+            this.renderSettingsForm();
+            
+        } catch (error) {
+            console.error('Error loading settings:', error);
+        }
+    }
+
+    renderSettingsForm() {
+        const form = document.getElementById('settings-form');
+        if (!form) return;
+        
+        // Server port
+        const portInput = document.getElementById('server-port');
+        if (portInput) portInput.value = this.settings.serverPort;
+        
+        // Log level
+        const logLevelSelect = document.getElementById('log-level');
+        if (logLevelSelect) logLevelSelect.value = this.settings.logLevel;
+        
+        // Database path
+        const dbPathInput = document.getElementById('db-path');
+        if (dbPathInput) dbPathInput.value = this.settings.dbPath;
+        
+        // Backup interval
+        const backupInput = document.getElementById('backup-interval');
+        if (backupInput) backupInput.value = this.settings.backupInterval;
+        
+        // Session timeout
+        const sessionInput = document.getElementById('session-timeout');
+        if (sessionInput) sessionInput.value = this.settings.sessionTimeout;
+    }
+
+    async loadSystemInfo() {
+        const container = document.getElementById('system-info');
+        if (!container) return;
+        
+        // In a real app, this would fetch system information
+        const systemInfo = {
+            version: '1.0.0',
+            uptime: '5 days, 3 hours',
+            platform: 'Linux x86_64',
+            node: 'v18.17.0',
+            memory: '2.4 GB / 8.0 GB',
+            disk: '45 GB / 256 GB',
+            lastBackup: '2024-01-15 02:00:00',
+            lastRestart: '2024-01-10 14:30:00'
+        };
+        
+        container.innerHTML = `
+            <div class="info-grid">
+                <div class="info-item">
+                    <span class="info-label">Version:</span>
+                    <span class="info-value">${systemInfo.version}</span>
+                </div>
+                <div class="info-item">
+                    <span class="info-label">Uptime:</span>
+                    <span class="info-value">${systemInfo.uptime}</span>
+                </div>
+                <div class="info-item">
+                    <span class="info-label">Platform:</span>
+                    <span class="info-value">${systemInfo.platform}</span>
+                </div>
+                <div class="info-item">
+                    <span class="info-label">Node.js:</span>
+                    <span class="info-value">${systemInfo.node}</span>
+                </div>
+                <div class="info-item">
+                    <span class="info-label">Memory:</span>
+                    <span class="info-value">${systemInfo.memory}</span>
+                </div>
+                <div class="info-item">
+                    <span class="info-label">Disk:</span>
+                    <span class="info-value">${systemInfo.disk}</span>
+                </div>
+                <div class="info-item">
+                    <span class="info-label">Last Backup:</span>
+                    <span class="info-value">${systemInfo.lastBackup}</span>
+                </div>
+                <div class="info-item">
+                    <span class="info-label">Last Restart:</span>
+                    <span class="info-value">${systemInfo.lastRestart}</span>
+                </div>
+            </div>
+        `;
+        
+        // Add CSS for info grid
+        this.addInfoStyles();
+    }
+
+    addInfoStyles() {
+        const style = document.createElement('style');
+        style.textContent = `
+            .info-grid {
+                display: grid;
+                grid-template-columns: repeat(auto-fill, minmax(250px, 1fr));
+                gap: 1rem;
+            }
+            
+            .info-item {
+                display: flex;
+                justify-content: space-between;
+                align-items: center;
+                padding: 0.75rem;
+                background-color: var(--bg-secondary);
+                border-radius: var(--border-radius-sm);
+            }
+            
+            .info-label {
+                font-size: 0.875rem;
+                color: var(--text-secondary);
+                font-weight: 500;
+            }
+            
+            .info-value {
+                font-size: 0.875rem;
+                color: var(--text-primary);
+                font-family: monospace;
+            }
+            
+            .form-section {
+                margin-bottom: 2rem;
+                padding-bottom: 1.5rem;
+                border-bottom: 1px solid var(--border-color);
+            }
+            
+            .form-section:last-child {
+                border-bottom: none;
+                margin-bottom: 0;
+            }
+            
+            .form-section h4 {
+                font-size: 1rem;
+                font-weight: 600;
+                color: var(--text-primary);
+                margin-bottom: 1rem;
+            }
+        `;
+        document.head.appendChild(style);
+    }
+
+    setupEventListeners() {
+        // Settings form
+        const form = document.getElementById('settings-form');
+        if (form) {
+            form.addEventListener('submit', (e) => {
+                e.preventDefault();
+                this.saveSettings();
+            });
+        }
+        
+        // Reset settings button
+        const resetBtn = document.getElementById('reset-settings');
+        if (resetBtn) {
+            resetBtn.addEventListener('click', () => {
+                this.resetSettings();
+            });
+        }
+        
+        // Database management buttons
+        const backupBtn = document.getElementById('backup-db');
+        if (backupBtn) {
+            backupBtn.addEventListener('click', () => {
+                this.backupDatabase();
+            });
+        }
+        
+        const optimizeBtn = document.getElementById('optimize-db');
+        if (optimizeBtn) {
+            optimizeBtn.addEventListener('click', () => {
+                this.optimizeDatabase();
+            });
+        }
+    }
+
+    saveSettings() {
+        // Collect form values
+        const settings = {
+            serverPort: parseInt(document.getElementById('server-port').value) || 8080,
+            logLevel: document.getElementById('log-level').value,
+            dbPath: document.getElementById('db-path').value,
+            backupInterval: parseInt(document.getElementById('backup-interval').value) || 24,
+            sessionTimeout: parseInt(document.getElementById('session-timeout').value) || 30,
+            dashboardPassword: document.getElementById('dashboard-password').value
+        };
+        
+        // Validate settings
+        if (settings.serverPort < 1024 || settings.serverPort > 65535) {
+            if (window.authManager) {
+                window.authManager.showToast('Server port must be between 1024 and 65535', 'error');
+            }
+            return;
+        }
+        
+        if (settings.backupInterval < 1 || settings.backupInterval > 168) {
+            if (window.authManager) {
+                window.authManager.showToast('Backup interval must be between 1 and 168 hours', 'error');
+            }
+            return;
+        }
+        
+        if (settings.sessionTimeout < 5 || settings.sessionTimeout > 1440) {
+            if (window.authManager) {
+                window.authManager.showToast('Session timeout must be between 5 and 1440 minutes', 'error');
+            }
+            return;
+        }
+        
+        // In a real app, this would save settings via API
+        this.settings = { ...this.settings, ...settings };
+        
+        if (window.authManager) {
+            window.authManager.showToast('Settings saved successfully', 'success');
+        }
+        
+        // Clear password field
+        document.getElementById('dashboard-password').value = '';
+    }
+
+    resetSettings() {
+        if (confirm('Are you sure you want to reset all settings to default values?')) {
+            // Reset to defaults
+            this.settings = {
+                serverPort: 8080,
+                logLevel: 'info',
+                dbPath: './data/llm-proxy.db',
+                backupInterval: 24,
+                sessionTimeout: 30,
+                enableRateLimiting: true,
+                enableCostTracking: true,
+                enableMetrics: true,
+                enableWebSocket: true
+            };
+            
+            this.renderSettingsForm();
+            
+            if (window.authManager) {
+                window.authManager.showToast('Settings reset to defaults', 'success');
+            }
+        }
+    }
+
+    backupDatabase() {
+        if (window.authManager) {
+            window.authManager.showToast('Starting database backup...', 'info');
+        }
+        
+        // Simulate backup process
+        setTimeout(() => {
+            // In a real app, this would trigger a database backup via API
+            if (window.authManager) {
+                window.authManager.showToast('Database backup completed successfully', 'success');
+            }
+        }, 2000);
+    }
+
+    optimizeDatabase() {
+        if (confirm('Optimize database? This may improve performance but could take a few moments.')) {
+            if (window.authManager) {
+                window.authManager.showToast('Optimizing database...', 'info');
+            }
+            
+            // Simulate optimization process
+            setTimeout(() => {
+                // In a real app, this would optimize the database via API
+                if (window.authManager) {
+                    window.authManager.showToast('Database optimization completed', 'success');
+                }
+            }, 3000);
+        }
+    }
+
+    refresh() {
+        this.loadSettings();
+        this.loadSystemInfo();
+    }
+}
+
+// Initialize settings page when needed
+window.initSettings = async () => {
+    window.settingsPage = new SettingsPage();
+};
+
+// Export for use in other modules
+if (typeof module !== 'undefined' && module.exports) {
+    module.exports = SettingsPage;
+}

static/js/websocket.js

@@ -0,0 +1,510 @@
+// WebSocket Manager for Real-time Updates
+
+class WebSocketManager {
+    constructor() {
+        this.ws = null;
+        this.reconnectAttempts = 0;
+        this.maxReconnectAttempts = 5;
+        this.reconnectDelay = 1000;
+        this.isConnected = false;
+        this.subscribers = new Map();
+        this.init();
+    }
+
+    init() {
+        this.connect();
+        this.setupStatusIndicator();
+        this.setupAutoReconnect();
+    }
+
+    connect() {
+        try {
+            // Determine WebSocket URL
+            const protocol = window.location.protocol === 'https:' ? 'wss:' : 'ws:';
+            const host = window.location.host;
+            const wsUrl = `${protocol}//${host}/ws`;
+            
+            this.ws = new WebSocket(wsUrl);
+            
+            this.ws.onopen = () => this.onOpen();
+            this.ws.onclose = () => this.onClose();
+            this.ws.onerror = (error) => this.onError(error);
+            this.ws.onmessage = (event) => this.onMessage(event);
+            
+        } catch (error) {
+            console.error('WebSocket connection error:', error);
+            this.scheduleReconnect();
+        }
+    }
+
+    onOpen() {
+        console.log('WebSocket connected');
+        this.isConnected = true;
+        this.reconnectAttempts = 0;
+        this.updateStatus('connected');
+        
+        // Notify subscribers
+        this.notify('connection', { status: 'connected' });
+        
+        // Send authentication if needed
+        if (window.authManager && window.authManager.token) {
+            this.send({
+                type: 'auth',
+                token: window.authManager.token
+            });
+        }
+        
+        // Subscribe to default channels
+        this.send({
+            type: 'subscribe',
+            channels: ['requests', 'metrics', 'logs']
+        });
+    }
+
+    onClose() {
+        console.log('WebSocket disconnected');
+        this.isConnected = false;
+        this.updateStatus('disconnected');
+        
+        // Notify subscribers
+        this.notify('connection', { status: 'disconnected' });
+        
+        // Schedule reconnection
+        this.scheduleReconnect();
+    }
+
+    onError(error) {
+        console.error('WebSocket error:', error);
+        this.updateStatus('error');
+    }
+
+    onMessage(event) {
+        try {
+            const data = JSON.parse(event.data);
+            this.handleMessage(data);
+        } catch (error) {
+            console.error('Error parsing WebSocket message:', error);
+        }
+    }
+
+    handleMessage(data) {
+        const { type, channel, payload } = data;
+        
+        // Notify channel subscribers
+        if (channel && this.subscribers.has(channel)) {
+            this.subscribers.get(channel).forEach(callback => {
+                try {
+                    callback(payload);
+                } catch (error) {
+                    console.error('Error in WebSocket callback:', error);
+                }
+            });
+        }
+        
+        // Handle specific message types
+        switch (type) {
+            case 'request':
+                this.handleRequest(payload);
+                break;
+            case 'metric':
+                this.handleMetric(payload);
+                break;
+            case 'log':
+                this.handleLog(payload);
+                break;
+            case 'system':
+                this.handleSystem(payload);
+                break;
+            case 'error':
+                this.handleError(payload);
+                break;
+        }
+    }
+
+    handleRequest(request) {
+        // Update request counters
+        this.updateRequestCounters(request);
+        
+        // Add to recent requests if on overview page
+        if (window.dashboard && window.dashboard.currentPage === 'overview') {
+            this.addRecentRequest(request);
+        }
+        
+        // Update monitoring stream if on monitoring page
+        if (window.dashboard && window.dashboard.currentPage === 'monitoring') {
+            this.addToMonitoringStream(request);
+        }
+    }
+
+    handleMetric(metric) {
+        // Update charts with new metric data
+        this.updateCharts(metric);
+        
+        // Update system metrics display
+        this.updateSystemMetrics(metric);
+    }
+
+    handleLog(log) {
+        // Add to logs table if on logs page
+        if (window.dashboard && window.dashboard.currentPage === 'logs') {
+            this.addLogEntry(log);
+        }
+        
+        // Add to monitoring logs if on monitoring page
+        if (window.dashboard && window.dashboard.currentPage === 'monitoring') {
+            this.addToLogStream(log);
+        }
+    }
+
+    handleSystem(system) {
+        // Update system health indicators
+        this.updateSystemHealth(system);
+    }
+
+    handleError(error) {
+        console.error('Server error:', error);
+        
+        // Show error toast
+        if (window.authManager) {
+            window.authManager.showToast(error.message || 'Server error', 'error');
+        }
+    }
+
+    send(data) {
+        if (this.ws && this.ws.readyState === WebSocket.OPEN) {
+            this.ws.send(JSON.stringify(data));
+            return true;
+        } else {
+            console.warn('WebSocket not connected, message not sent:', data);
+            return false;
+        }
+    }
+
+    subscribe(channel, callback) {
+        if (!this.subscribers.has(channel)) {
+            this.subscribers.set(channel, new Set());
+        }
+        this.subscribers.get(channel).add(callback);
+        
+        // Send subscription to server
+        this.send({
+            type: 'subscribe',
+            channels: [channel]
+        });
+        
+        // Return unsubscribe function
+        return () => this.unsubscribe(channel, callback);
+    }
+
+    unsubscribe(channel, callback) {
+        if (this.subscribers.has(channel)) {
+            this.subscribers.get(channel).delete(callback);
+            
+            // If no more subscribers, unsubscribe from server
+            if (this.subscribers.get(channel).size === 0) {
+                this.send({
+                    type: 'unsubscribe',
+                    channels: [channel]
+                });
+            }
+        }
+    }
+
+    notify(channel, data) {
+        if (this.subscribers.has(channel)) {
+            this.subscribers.get(channel).forEach(callback => {
+                try {
+                    callback(data);
+                } catch (error) {
+                    console.error('Error in notification callback:', error);
+                }
+            });
+        }
+    }
+
+    scheduleReconnect() {
+        if (this.reconnectAttempts >= this.maxReconnectAttempts) {
+            console.warn('Max reconnection attempts reached');
+            return;
+        }
+        
+        this.reconnectAttempts++;
+        const delay = this.reconnectDelay * Math.pow(2, this.reconnectAttempts - 1);
+        
+        console.log(`Scheduling reconnection in ${delay}ms (attempt ${this.reconnectAttempts})`);
+        
+        setTimeout(() => {
+            if (!this.isConnected) {
+                this.connect();
+            }
+        }, delay);
+    }
+
+    setupAutoReconnect() {
+        // Reconnect when browser comes online
+        window.addEventListener('online', () => {
+            if (!this.isConnected) {
+                console.log('Browser online, attempting to reconnect...');
+                this.connect();
+            }
+        });
+        
+        // Keepalive ping
+        setInterval(() => {
+            if (this.isConnected) {
+                this.send({ type: 'ping' });
+            }
+        }, 30000);
+    }
+
+    setupStatusIndicator() {
+        // Status indicator is already in the HTML
+        // This function just ensures it's properly styled
+    }
+
+    updateStatus(status) {
+        const statusElement = document.getElementById('ws-status');
+        if (!statusElement) return;
+        
+        const dot = statusElement.querySelector('.ws-dot');
+        const text = statusElement.querySelector('.ws-text');
+        
+        if (!dot || !text) return;
+        
+        // Remove all status classes
+        dot.classList.remove('connected', 'disconnected');
+        statusElement.classList.remove('connected', 'disconnected');
+        
+        // Add new status class
+        dot.classList.add(status);
+        statusElement.classList.add(status);
+        
+        // Update text
+        const statusText = {
+            'connected': 'Connected',
+            'disconnected': 'Disconnected',
+            'connecting': 'Connecting...',
+            'error': 'Connection Error'
+        };
+        
+        text.textContent = statusText[status] || status;
+    }
+
+    // Helper methods for updating UI
+    updateRequestCounters(request) {
+        // Update request counters in overview stats
+        const requestCountElement = document.querySelector('[data-stat="total-requests"]');
+        if (requestCountElement) {
+            const currentCount = parseInt(requestCountElement.textContent) || 0;
+            requestCountElement.textContent = currentCount + 1;
+        }
+        
+        // Update token counters
+        const tokenCountElement = document.querySelector('[data-stat="total-tokens"]');
+        if (tokenCountElement && request.tokens) {
+            const currentTokens = parseInt(tokenCountElement.textContent) || 0;
+            tokenCountElement.textContent = currentTokens + request.tokens;
+        }
+    }
+
+    addRecentRequest(request) {
+        const tableBody = document.querySelector('#recent-requests tbody');
+        if (!tableBody) return;
+        
+        const row = document.createElement('tr');
+        
+        // Format time
+        const time = new Date(request.timestamp || Date.now()).toLocaleTimeString();
+        
+        // Format status badge
+        const statusClass = request.status === 'success' ? 'success' : 
+                          request.status === 'error' ? 'danger' : 'warning';
+        const statusIcon = request.status === 'success' ? 'check-circle' :
+                          request.status === 'error' ? 'exclamation-circle' : 'exclamation-triangle';
+        
+        row.innerHTML = `
+            <td>${time}</td>
+            <td>${request.client_id || 'Unknown'}</td>
+            <td>${request.provider || 'Unknown'}</td>
+            <td>${request.model || 'Unknown'}</td>
+            <td>${request.tokens || 0}</td>
+            <td>
+                <span class="status-badge ${statusClass}">
+                    <i class="fas fa-${statusIcon}"></i>
+                    ${request.status || 'unknown'}
+                </span>
+            </td>
+        `;
+        
+        // Add to top of table
+        tableBody.insertBefore(row, tableBody.firstChild);
+        
+        // Limit to 50 rows
+        const rows = tableBody.querySelectorAll('tr');
+        if (rows.length > 50) {
+            tableBody.removeChild(rows[rows.length - 1]);
+        }
+    }
+
+    addToMonitoringStream(request) {
+        const streamElement = document.getElementById('request-stream');
+        if (!streamElement) return;
+        
+        const entry = document.createElement('div');
+        entry.className = 'stream-entry';
+        
+        // Format time
+        const time = new Date().toLocaleTimeString();
+        
+        // Determine icon based on status
+        let icon = 'question-circle';
+        let color = 'var(--text-secondary)';
+        
+        if (request.status === 'success') {
+            icon = 'check-circle';
+            color = 'var(--success)';
+        } else if (request.status === 'error') {
+            icon = 'exclamation-circle';
+            color = 'var(--danger)';
+        }
+        
+        entry.innerHTML = `
+            <div class="stream-entry-time">${time}</div>
+            <div class="stream-entry-icon" style="color: ${color}">
+                <i class="fas fa-${icon}"></i>
+            </div>
+            <div class="stream-entry-content">
+                <strong>${request.client_id || 'Unknown'}</strong> → 
+                ${request.provider || 'Unknown'} (${request.model || 'Unknown'})
+                <div class="stream-entry-details">
+                    ${request.tokens || 0} tokens • ${request.duration || 0}ms
+                </div>
+            </div>
+        `;
+        
+        // Add to top of stream
+        streamElement.insertBefore(entry, streamElement.firstChild);
+        
+        // Limit to 20 entries
+        const entries = streamElement.querySelectorAll('.stream-entry');
+        if (entries.length > 20) {
+            streamElement.removeChild(entries[entries.length - 1]);
+        }
+        
+        // Add highlight animation
+        entry.classList.add('highlight');
+        setTimeout(() => entry.classList.remove('highlight'), 1000);
+    }
+
+    updateCharts(metric) {
+        // This would update Chart.js charts with new data
+        // Implementation depends on specific chart setup
+    }
+
+    updateSystemMetrics(metric) {
+        const metricsElement = document.getElementById('system-metrics');
+        if (!metricsElement) return;
+        
+        // Update specific metric displays
+        // This is a simplified example
+    }
+
+    addLogEntry(log) {
+        const tableBody = document.querySelector('#logs-table tbody');
+        if (!tableBody) return;
+        
+        const row = document.createElement('tr');
+        
+        // Format time
+        const time = new Date(log.timestamp || Date.now()).toLocaleString();
+        
+        // Determine log level class
+        const levelClass = log.level || 'info';
+        
+        row.innerHTML = `
+            <td>${time}</td>
+            <td>
+                <span class="status-badge ${levelClass}">
+                    ${levelClass.toUpperCase()}
+                </span>
+            </td>
+            <td>${log.source || 'Unknown'}</td>
+            <td>${log.message || ''}</td>
+        `;
+        
+        // Add to top of table
+        tableBody.insertBefore(row, tableBody.firstChild);
+        
+        // Limit to 100 rows
+        const rows = tableBody.querySelectorAll('tr');
+        if (rows.length > 100) {
+            tableBody.removeChild(rows[rows.length - 1]);
+        }
+    }
+
+    addToLogStream(log) {
+        const logStreamElement = document.getElementById('system-logs');
+        if (!logStreamElement) return;
+        
+        const entry = document.createElement('div');
+        entry.className = `log-entry log-${log.level || 'info'}`;
+        
+        // Format time
+        const time = new Date().toLocaleTimeString();
+        
+        // Determine icon based on level
+        let icon = 'info-circle';
+        if (log.level === 'error') icon = 'exclamation-circle';
+        if (log.level === 'warn') icon = 'exclamation-triangle';
+        if (log.level === 'debug') icon = 'bug';
+        
+        entry.innerHTML = `
+            <div class="log-time">${time}</div>
+            <div class="log-level">
+                <i class="fas fa-${icon}"></i>
+            </div>
+            <div class="log-message">${log.message || ''}</div>
+        `;
+        
+        // Add to top of stream
+        logStreamElement.insertBefore(entry, logStreamElement.firstChild);
+        
+        // Limit to 50 entries
+        const entries = logStreamElement.querySelectorAll('.log-entry');
+        if (entries.length > 50) {
+            logStreamElement.removeChild(entries[entries.length - 1]);
+        }
+    }
+
+    updateSystemHealth(system) {
+        const healthElement = document.getElementById('system-health');
+        if (!healthElement) return;
+        
+        // Update system health indicators
+        // This is a simplified example
+    }
+
+    disconnect() {
+        if (this.ws) {
+            this.ws.close();
+            this.ws = null;
+        }
+        this.isConnected = false;
+        this.updateStatus('disconnected');
+    }
+
+    reconnect() {
+        this.disconnect();
+        this.connect();
+    }
+}
+
+// Initialize WebSocket manager when DOM is loaded
+document.addEventListener('DOMContentLoaded', () => {
+    window.wsManager = new WebSocketManager();
+});
+
+// Export for use in other modules
+if (typeof module !== 'undefined' && module.exports) {
+    module.exports = WebSocketManager;
+}

test_dashboard.sh

@@ -0,0 +1,38 @@
+#!/bin/bash
+
+# Test script for LLM Proxy Dashboard
+
+echo "Building LLM Proxy Gateway..."
+cargo build --release
+
+echo ""
+echo "Starting server in background..."
+./target/release/llm-proxy &
+SERVER_PID=$!
+
+# Wait for server to start
+sleep 3
+
+echo ""
+echo "Testing dashboard endpoints..."
+
+# Test health endpoint
+echo "1. Testing health endpoint:"
+curl -s http://localhost:8080/health
+
+echo ""
+echo "2. Testing dashboard static files:"
+curl -s -o /dev/null -w "%{http_code}" http://localhost:8080/
+
+echo ""
+echo "3. Testing API endpoints:"
+curl -s http://localhost:8080/api/auth/status | jq . 2>/dev/null || echo "JSON response received"
+
+echo ""
+echo "Dashboard should be available at: http://localhost:8080"
+echo "Default login: admin / admin123"
+echo ""
+echo "Press Ctrl+C to stop the server"
+
+# Keep script running
+wait $SERVER_PID

test_server.sh

@@ -0,0 +1,75 @@
+#!/bin/bash
+
+# Test script for LLM Proxy Gateway
+
+echo "Building LLM Proxy Gateway..."
+cargo build --release
+
+if [ $? -ne 0 ]; then
+    echo "Build failed!"
+    exit 1
+fi
+
+echo "Build successful!"
+
+echo ""
+echo "Project Structure Summary:"
+echo "=========================="
+echo "Core Components:"
+echo "  - main.rs: Application entry point with server setup"
+echo "  - config/: Configuration management"
+echo "  - server/: API route handlers"
+echo "  - auth/: Bearer token authentication"
+echo "  - database/: SQLite database setup"
+echo "  - models/: Data structures (OpenAI-compatible)"
+echo "  - providers/: LLM provider implementations (OpenAI, Gemini, DeepSeek, Grok)"
+echo "  - errors/: Custom error types"
+echo "  - dashboard/: Admin dashboard with WebSocket support"
+echo "  - logging/: Request logging middleware"
+echo "  - state/: Shared application state"
+echo "  - multimodal/: Image processing support (basic structure)"
+
+echo ""
+echo "Key Features Implemented:"
+echo "=========================="
+echo "✓ OpenAI-compatible API endpoint (/v1/chat/completions)"
+echo "✓ Bearer token authentication"
+echo "✓ SQLite database for request tracking"
+echo "✓ Request logging with token/cost calculation"
+echo "✓ Provider abstraction layer"
+echo "✓ Admin dashboard with real-time monitoring"
+echo "✓ WebSocket support for live updates"
+echo "✓ Configuration management (config.toml, .env, env vars)"
+echo "✓ Multimodal support structure (images)"
+echo "✓ Error handling with proper HTTP status codes"
+
+echo ""
+echo "Next Steps Needed:"
+echo "=================="
+echo "1. Add API keys to .env file:"
+echo "   OPENAI_API_KEY=your_key_here"
+echo "   GEMINI_API_KEY=your_key_here"
+echo "   DEEPSEEK_API_KEY=your_key_here"
+echo "   GROK_API_KEY=your_key_here (optional)"
+echo ""
+echo "2. Create config.toml for custom configuration (optional)"
+echo ""
+echo "3. Run the server:"
+echo "   cargo run"
+echo ""
+echo "4. Access dashboard at: http://localhost:8080"
+echo ""
+echo "5. Test API with curl:"
+echo "   curl -X POST http://localhost:8080/v1/chat/completions \\"
+echo "     -H 'Authorization: Bearer your_token' \\"
+echo "     -H 'Content-Type: application/json' \\"
+echo "     -d '{\"model\": \"gpt-4\", \"messages\": [{\"role\": \"user\", \"content\": \"Hello\"}]}'"
+
+echo ""
+echo "Deployment Notes:"
+echo "================="
+echo "Memory: Designed for 512MB RAM (LXC container)"
+echo "Database: SQLite (./data/llm_proxy.db)"
+echo "Port: 8080 (configurable)"
+echo "Authentication: Single Bearer token (configurable)"
+echo "Providers: OpenAI, Gemini, DeepSeek, Grok (disabled by default)"

tests/integration_tests.rs.bak

@@ -0,0 +1,188 @@
+// Integration tests for LLM Proxy Gateway
+
+use llm_proxy::config::Config;
+use llm_proxy::database::Database;
+use llm_proxy::state::AppState;
+use llm_proxy::rate_limiting::RateLimitManager;
+use tempfile::TempDir;
+use std::fs;
+
+#[tokio::test]
+async fn test_config_loading() {
+    // Create a temporary config file
+    let temp_dir = TempDir::new().unwrap();
+    let config_path = temp_dir.path().join("config.toml");
+    
+    let config_content = r#"
+[server]
+port = 8080
+host = "0.0.0.0"
+
+[database]
+path = "./data/test.db"
+max_connections = 5
+
+[providers.openai]
+enabled = true
+base_url = "https://api.openai.com/v1"
+
+[providers.gemini]
+enabled = true
+base_url = "https://generativelanguage.googleapis.com/v1"
+
+[providers.deepseek]
+enabled = true
+base_url = "https://api.deepseek.com"
+
+[providers.grok]
+enabled = false
+base_url = "https://api.x.ai/v1"
+
+[model_mapping]
+"gpt-*" = "openai"
+"gemini-*" = "gemini"
+"deepseek-*" = "deepseek"
+"grok-*" = "grok"
+
+[pricing]
+openai = { input = 0.01, output = 0.03 }
+gemini = { input = 0.0005, output = 0.0015 }
+deepseek = { input = 0.00014, output = 0.00028 }
+grok = { input = 0.001, output = 0.003 }
+"#;
+    
+    fs::write(&config_path, config_content).unwrap();
+    
+    // Test loading config
+    let config = Config::load_from_path(&config_path);
+    assert!(config.is_ok());
+    
+    let config = config.unwrap();
+    assert_eq!(config.server.port, 8080);
+    assert!(config.providers.openai.is_some());
+    assert!(config.providers.grok.is_none());
+}
+
+#[tokio::test]
+async fn test_database_initialization() {
+    // Create a temporary database file
+    let temp_dir = TempDir::new().unwrap();
+    let db_path = temp_dir.path().join("test.db");
+    
+    // Test database initialization
+    let database = Database::new(&db_path).await;
+    assert!(database.is_ok());
+    
+    let database = database.unwrap();
+    
+    // Test connection
+    let test_result = database.test_connection().await;
+    assert!(test_result.is_ok());
+}
+
+#[tokio::test]
+async fn test_provider_manager() {
+    // Create a provider manager
+    use llm_proxy::providers::{ProviderManager, Provider};
+    use llm_proxy::config::OpenAIConfig;
+    
+    let mut manager = ProviderManager::new();
+    assert_eq!(manager.providers.len(), 0);
+    
+    // Test adding providers (we can't actually add real providers without API keys)
+    // This test just verifies the manager structure works
+    assert!(manager.get_provider_for_model("gpt-4").is_none());
+    assert!(manager.get_provider("openai").is_none());
+}
+
+#[tokio::test]
+async fn test_rate_limit_manager() {
+    let manager = RateLimitManager::new(60, 10);
+    
+    // Test client rate limiting
+    let allowed = manager.check_request("test-client").await;
+    assert!(allowed); // First request should be allowed
+    
+    // Test provider circuit breaker
+    let allowed = manager.check_provider("openai").await;
+    assert!(allowed); // Circuit should be closed initially
+    
+    // Record some failures
+    manager.record_provider_failure("openai").await;
+    manager.record_provider_failure("openai").await;
+    manager.record_provider_failure("openai").await;
+    manager.record_provider_failure("openai").await;
+    manager.record_provider_failure("openai").await;
+    
+    // After 5 failures, circuit should be open
+    let allowed = manager.check_provider("openai").await;
+    assert!(!allowed); // Circuit should be open
+    
+    // Record success to close circuit
+    manager.record_provider_success("openai").await;
+    manager.record_provider_success("openai").await;
+    manager.record_provider_success("openai").await;
+    
+    // After 3 successes in half-open state, circuit should be closed
+    let allowed = manager.check_provider("openai").await;
+    assert!(allowed); // Circuit should be closed again
+}
+
+#[tokio::test]
+async fn test_app_state_creation() {
+    // Create a temporary database
+    let temp_dir = TempDir::new().unwrap();
+    let db_path = temp_dir.path().join("test.db");
+    
+    let database = Database::new(&db_path).await.unwrap();
+    
+    // Test AppState creation using test utilities
+    use llm_proxy::test_utils::create_test_state;
+    let state = create_test_state().await;
+    
+    // Verify state components are initialized
+    assert!(state.database.test_connection().await.is_ok());
+}
+
+#[tokio::test]
+async fn test_multimodal_image_converter() {
+    use llm_proxy::multimodal::{ImageConverter, ImageInput};
+    
+    // Test model detection
+    assert!(ImageConverter::model_supports_multimodal("gpt-4-vision-preview"));
+    assert!(ImageConverter::model_supports_multimodal("gemini-pro-vision"));
+    assert!(!ImageConverter::model_supports_multimodal("gpt-3.5-turbo"));
+    assert!(!ImageConverter::model_supports_multimodal("gemini-pro"));
+    
+    // Test data URL parsing (utility function)
+    let test_url = "data:image/jpeg;base64,SGVsbG8gV29ybGQ=";
+    let parts: Vec<&str> = test_url[5..].split(";base64,").collect();
+    assert_eq!(parts.len(), 2);
+    assert_eq!(parts[0], "image/jpeg");
+    assert_eq!(parts[1], "SGVsbG8gV29ybGQ=");
+}
+
+#[tokio::test]
+async fn test_error_conversions() {
+    use llm_proxy::errors::AppError;
+    use anyhow::anyhow;
+    
+    // Test anyhow error conversion
+    let anyhow_error = anyhow!("Test error");
+    let app_error: AppError = anyhow_error.into();
+    
+    match app_error {
+        AppError::InternalError(msg) => assert_eq!(msg, "Test error"),
+        _ => panic!("Expected InternalError"),
+    }
+    
+    // Test sqlx error conversion
+    use sqlx::Error as SqlxError;
+    let sqlx_error = SqlxError::PoolClosed;
+    let app_error: AppError = sqlx_error.into();
+    
+    match app_error {
+        AppError::DatabaseError(msg) => assert!(msg.contains("pool closed")),
+        _ => panic!("Expected DatabaseError"),
+    }
+}