Unified AI Router

Appearance

Quickstart ​

Install, configure, and test in 5 minutes.

Install ​

bash
git clone https://github.com/mlibre/Unified-AI-Router.git
cd Unified-AI-Router
npm install

Configure ​

bash
cp .env.example .env
# Edit .env and add OPENROUTER_API_KEY or other provider keys

Edit provider.js to set your providers.

Run ​

bash
npm start

Server starts on http://localhost:3000 with these endpoints:

  • POST /v1/responses - OpenAI Responses API
  • POST /v1/chat/completions - Chat Completions API
  • GET /v1/models - List models
  • GET /health - Health check

Test ​

Chat Completions:

bash
curl -X POST http://localhost:3000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{"messages": [{"role":"user","content":"Hello"}], "model":"no_need"}'

Responses API:

bash
curl -X POST http://localhost:3000/v1/responses \
  -H "Content-Type: application/json" \
  -d '{"input": "Hello", "model":"no_need"}'

Streaming:

bash
curl -X POST http://localhost:3000/v1/responses \
  -H "Content-Type: application/json" \
  -d '{"input": "Say hello in 3 words", "model":"any-model", "stream": true}'

Next ​

  • Configure more providers in provider.js
  • Check provider status: GET /providers/status
  • See Testing Guide for comprehensive testing
  • Check out SDK Usage for library integration