OpenAI-Compatible Fal.ai Proxy

A high-performance Deno proxy that makes Fal.ai's powerful models available through the standard OpenAI /v1/images/generations API.

中文版

✨ Key Features

• 🔌 Drop-in Compatibility: Use Fal.ai models with any existing OpenAI-compatible client or library.
• ⚙️ Configuration-Driven: All settings, including the list of supported models, are managed in a simple .env file. No code changes needed.
• 🧠 Dynamic Model Adaptation: Automatically fetches and analyzes each model's OpenAPI schema to intelligently map parameters like size, width/height, and aspect_ratio.
• ⚡ High Performance: Features an on-startup cache for model schemas, reducing API latency for all subsequent requests.
• 🔐 Centralized API Key Management: Securely manage your Fal.ai keys on the server and provide a single, custom access key to your clients.
• 🌐 CORS Ready: Built-in CORS support allows direct access from web applications.

🚀 Quick Start

1. Get the code

Download router.ts or clone the repository.

2. Create your configuration

Create a .env file in the same directory and populate it with your keys and desired models.

.env file example:

# Your secret key to access THIS proxy
CUSTOM_ACCESS_KEY="my-super-secret-proxy-key"

# A comma-separated list of your Fal.ai API keys
AI_KEYS="fal-key-123abc,fal-key-456def"

# (Optional) Port to run the server on
PORT="8000"

# (Optional) Enable detailed console logs
DEBUG_MODE="true"

# Define the models you want to expose
# Format: "friendly-name:fal-ai/endpoint/id,another-name:another/endpoint"
SUPPORTED_MODELS="flux-dev:fal-ai/flux/dev,sdxl:fal-ai/stable-diffusion-xl,flux-schnell:fal-ai/flux-schnell"

3. Run the server

Start the Deno process with the necessary permissions.

deno run --allow-net --allow-read=.env --allow-env router.ts

The server will start, pre-load all model configurations, and be ready to accept requests.

🎯 Usage (API Endpoints)

Generating an Image

Send a POST request to /v1/images/generations. The proxy will translate it and forward it to the appropriate Fal.ai model.

Example with curl:

curl -X POST http://localhost:8000/v1/images/generations \
  -H "Authorization: Bearer my-super-secret-proxy-key" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "A majestic dragon soaring through clouds, cinematic lighting",
    "model": "flux-dev",
    "size": "1024x1024",
    "n": 1
  }'

Other Endpoints

• List Models: GET /v1/models - Returns a list of all models configured in your .env file, formatted like the OpenAI models API.
• Health Check: GET /health - A simple endpoint that returns { "status": "ok" } for monitoring.

⚙️ Configuration Details

All configuration is managed via the .env file.