GitHub radar
Free AI Gateway for Claude Code and Cursor: OmniRoute
A local proxy that routes your Claude Code, Cursor, or Cline requests to 352 different AI providers, with automatic failover and ~1.51 billion free tokens / month via free tiers.
OmniRoute is a self-hosted AI gateway that acts as a proxy between your code editor (Claude Code, Cursor, Cline, Copilot, Codex) and AI model providers. It provides an OpenAI compatible endpoint on localhost:20128 and automatically routes your requests to 352 providers & 1,200+ models with automatic failover when quotas are exceeded. With no API keys it uses keyless free providers, and with API keys you can stack up to ~1.51 billion free tokens / month with free tier API keys. It also includes a token compression layer (RTK + Caveman) that compresses the resulting bash commands sent to the model by 15–95% to further reduce costs. An npm package, Docker image, desktop app, and PWA are available. Also supports MCP & A2A. Over 600 contributors, includes a Russian language README.
Why a vibe-coder should care
Self funding: Instead of buying a premium service and hitting rate limits during your session, automatically fallback to hundreds of providers, including ~1.51 billion free tokens / month. Just one config change in Claude Code or Cursor and you’re good to go!
How to install
Copy this and send it to your agent — Claude Code, Codex, any of them:
Install OmniRoute from https://github.com/diegosouzapw/OmniRoute — follow the README (npm i -g omniroute starts the gateway on localhost:20128), then configure Claude Code to use that address as the base URL for model requests. Show me that a test call goes through successfully.
Runs on a regular laptop — only Node.js or Docker needed.
Open on GitHub▌ More finds