OpenRouter Bridge

OpenAI-compatible proxy server that routes LLM requests through OpenRouter for low-cost inference. Built for Nomad Internet's Lyzr AI agent stack.

What It Does

• Exposes /v1/chat/completions — drop-in replacement for OpenAI API
• Exposes /v1/models — lists all available OpenRouter models
• Model aliasing — automatically maps expensive model names (gpt-4o, claude-3-opus) to cheap equivalents (Mistral Small, Gemini Flash)
• Supports streaming responses
• CORS enabled for browser-based integrations

Model Alias Map

You Request	Actually Runs	Cost/1M
gpt-4o	google/gemini-2.0-flash-001	$0.10
gpt-4o-mini	mistralai/mistral-small-3.1-24b-instruct	$0.03
claude-3-5-sonnet-latest	google/gemini-2.0-flash-001	$0.10
claude-3-opus-20240229	qwen/qwq-32b	$0.15
gemini-2.5-flash	google/gemini-2.5-flash-preview:thinking	$0.30

Deploy to Railway

1. Push this repo to GitHub
2. Create a new Railway project from the repo
3. Set environment variable: OPENROUTER_API_KEY=sk-or-v1-...
4. Deploy — Railway auto-detects Python and uses the Procfile

Usage

# Test the bridge
curl https://your-bridge.railway.app/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o-mini",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Use with Any OpenAI-Compatible Tool

Set base URL to: https://your-bridge.railway.app/v1 No API key required (the bridge holds the OpenRouter key server-side).