Features How It Works FAQ GitHub
Open Source · Free to use

Your API Gateway Lightweight & Fast

A lightweight OpenAI-compatible API gateway featuring multi-key polling, failover, multi-provider routing, custom route mapping, key management, and a simple dashboard. Runs on bare Node.js with zero dependencies.

Download GitHub
API Gateway Console Dashboard Screenshot

Easy config and monitoring dashboard

Everything You Need

Zero dependencies. Deploy anywhere in seconds.

Multi-Provider Routing

Route to OpenAI, DeepSeek, Mistral, Groq, Anthropic, VyceAI, AMD Radeon, BazaarLink, FlatKey, TokenRouter, or any OpenAI-compatible provider from a single endpoint.

Key Management

A simple JSON-formatted key list for easy configuration and backup.

Easy Dashboard

Metrics: request rate, latency, error codes, provider health. Built-in console served from memory.

Log Analytics

Track request usage and errors.

Provider Health Probes

Continuous health checks with visual status indicators. Automatic provider degradation detection.

Rate Limit

Support per-key RPM and model context size limits.

Quick Start

Need Node v18+ or Bun JS runtime, download the repository, run the NPM script.

  1. 1
    Download repository for your platform.
  2. 2
    Run npm start
    The dashboard is at http://localhost:3000/console
    That's it. 🎉
  3. 3
    Configure client auth key (it's login password too), providers, keys, and routes in config page.
  4. 4
    Tty default is to use a free upstream API, For testing purposes only, recommended to replace it with your own stable upstream source.
// settings panel or manually the config.json
{
  "port": 3000,
  "providers": {
    "openai": {
      "apiKeys": ["sk-...KeyA", "sk-...KeyB"],
      "baseUrl": "https://api.openai.com"
    }
  },
  "models": {
    "gpt-5.6": "openai"
  }
}

FAQ

Quick answers to common questions.

Is everything free?
This project is free and open-source, but whether your API keys incur charges depends on your provider.
What providers are supported?
OpenAI, DeepSeek, Mistral, Groq, Anthropic, VyceAI, Navy, Ollama, AMD Radeon, BazaarLink, FlatKey, TokenRouter, Cloudflare Workers AI, and any OpenAI-compatible endpoint. Add custom providers via the config file — no code changes needed.
Do I need a database?
No. Everything runs in-memory with optional file-based persistence for stats. No PostgreSQL, no Redis, no external dependencies.
Can I run it behind a reverse proxy?
Yes. The dashboard runs on a configurable base path (e.g. http://your.host/console). Works with nginx, Caddy, Cloudflare Tunnel, or any reverse proxy.

More details

Star on GitHub, download the repository, or contribute.

Download GitHub

Loading...

Loading...