README.md

metadata

title: E2B API Proxy
emoji: 🚀
colorFrom: blue
colorTo: indigo
sdk: docker
pinned: false
app_port: 7860

E2B API Proxy with FastAPI

This project is a FastAPI implementation of an API proxy for E2B (fragments.e2b.dev). It provides a compatible interface for various AI model providers including OpenAI, Google, and Anthropic.

Description

The E2B API Proxy acts as a middleware between your application and the E2B service, providing:

• Proxy API requests to E2B service
• Support for multiple AI models (OpenAI, Google Vertex AI, Anthropic)
• Streaming and non-streaming response handling
• CORS support for cross-origin requests

Deployment on Hugging Face Spaces

This application is ready to be deployed on Hugging Face Spaces:

1. Create a new Space on Hugging Face with Docker SDK
2. Upload these files to your Space
3. Set the environment variables in the Space settings:
   • API_KEY: Your API key for authentication (default: sk-123456)
   • API_BASE_URL: The base URL for the E2B service (default: https://fragments.e2b.dev)

API Endpoints

• GET /hf/v1/models: List available models
• POST /hf/v1/chat/completions: Send chat completion requests
• GET /: Root endpoint (health check)

Configuration

The main configuration is in the app.py file. You can customize:

• API key for authentication
• Base URL for the E2B service
• Model configurations
• Default headers for requests

Local Development

Prerequisites

• Docker and Docker Compose

Running the Application Locally

1. Clone this repository
2. Update the API key in docker-compose.yml (replace sk-123456 with your actual key)
3. Build and start the container:

docker-compose up -d

4. The API will be available at http://localhost:7860

Testing the API

You can test the API using curl:

# Get available models
curl http://localhost:7860/hf/v1/models

# Send a chat completion request
curl -X POST http://localhost:7860/hf/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-123456" \
  -d '{
    "model": "gpt-4o",
    "messages": [
      {"role": "user", "content": "Hello, how are you?"}
    ]
  }'

Supported Models

The API supports various models from different providers:

• OpenAI: o1-preview, o3-mini, gpt-4o, gpt-4.5-preview, gpt-4-turbo
• Google: gemini-1.5-pro, gemini-2.5-pro-exp-03-25, gemini-exp-1121, gemini-2.0-flash-exp
• Anthropic: claude-3-5-sonnet-latest, claude-3-7-sonnet-latest, claude-3-5-haiku-latest

License

This project is open source and available under the MIT License.