obscuraaihub 1.0.0

Multi-modal AI API Gateway - Unified interface for multiple AI providers

Obscura API Hub - FastAPI Backend

Multi-modal AI API Gateway with intelligent routing and provider management.

Features

• 🔐 Token-based authentication
• 🎯 Intelligent routing strategies (auto/cost/speed/quality/balanced)
• 🔄 Automatic API key rotation
• 📊 Usage tracking and quotas
• 🚀 Multi-modal support (text, image, video, audio)
• 🔌 OpenAI-compatible API
• 💪 Provider health monitoring
• ⚡ Streaming support

Setup

1. Install dependencies:

pip install -r requirements.txt

2. Configure environment:

cp .env.example .env
# Edit .env with your configuration

3. Run the server:

# Development
python -m app.main

# Production
uvicorn app.main:app --host 0.0.0.0 --port 8000

API Endpoints

OpenAI-Compatible Endpoints

• POST /v1/chat/completions - Chat completions with multimodal support
• POST /v1/embeddings - Text embeddings
• GET /v1/models - List available models
• POST /v1/audio/transcriptions - Audio transcription (Whisper)
• POST /v1/images/generations - Image generation (Lovable AI)
• POST /v1/files - File upload to R2

Internal Endpoints

• GET /health - Health check
• POST /internal/models/refresh - Refresh model cache

Multimodal Support ⭐

Supports vision, video, and audio understanding through SiliconFlow:

# Vision understanding
{
    "model": "Qwen/Qwen2.5-VL-72B-Instruct",
    "messages": [{
        "role": "user",
        "content": [
            {"type": "image_url", "image_url": {"url": "...", "detail": "high"}},
            {"type": "text", "text": "描述这张图片"}
        ]
    }]
}

# Video understanding
{
    "model": "Qwen/Qwen3-Omni-30B-A3B-Instruct",
    "messages": [{
        "role": "user",
        "content": [
            {"type": "video_url", "video_url": {"url": "...", "max_frames": 16, "fps": 1}},
            {"type": "text", "text": "总结视频内容"}
        ]
    }]
}

# Audio understanding
{
    "model": "Qwen/Qwen3-Omni-30B-A3B-Instruct",
    "messages": [{
        "role": "user",
        "content": [
            {"type": "audio_url", "audio_url": {"url": "..."}},
            {"type": "text", "text": "转录这段音频"}
        ]
    }]
}

API Documentation

Once running, visit:

• Swagger UI: http://localhost:8000/docs
• ReDoc: http://localhost:8000/redoc

Project Structure

backend/
├── app/
│   ├── api/              # API endpoints
│   │   ├── v1/           # v1 endpoints (OpenAI-compatible)
│   │   └── internal/     # Internal endpoints
│   ├── core/             # Core functionality (auth, routing)
│   ├── db/               # Database operations
│   ├── models/           # Pydantic models
│   ├── providers/        # LLM provider implementations
│   ├── services/         # Business logic services
│   ├── utils/            # Utilities
│   └── main.py           # FastAPI app
├── tests/                # Tests
└── requirements.txt      # Dependencies

Testing

pytest tests/

Environment Variables

See .env.example for all configuration options.

Key Features

Intelligent Routing

• auto: Balance health and priority
• cost: Minimize cost
• speed: Minimize latency
• quality: Maximize quality
• balanced: Balance all factors

Provider Health Monitoring

• Automatic health tracking
• Failure rate calculation
• Unhealthy provider detection (>30% failure rate)

API Key Rotation

• Round-robin rotation
• Health-based selection
• Automatic failover

Model Binding

• Models can be bound to specific providers
• Cross-provider fallback disabled for bound models
• Automatic model discovery and caching

License

MIT