OmniRoute is a free, MIT-licensed AI gateway that runs on your local machine, routing requests to 237 AI providers (90+ with free tiers) and providing automatic fallback so you never hit rate limits.
What is OmniRoute?
OmniRoute is a free, MIT-licensed AI gateway that runs entirely on your own machine. It takes API requests from coding tools and routes them to 237 AI providers, 90+ of which offer free tiers. It outputs responses from the selected provider, with automatic fallback when a provider hits a rate limit. It runs locally as a server (npm install) and is maintained by the open-source community on GitHub.
Key Features
- Multi-provider aggregation — OmniRoute connects to 237 AI providers, including 90+ free tiers, aggregating roughly 1.6 billion free tokens per month.
- Auto-fallback routing — 4-tier fallback chain (Subscription → API → Cheap → Free) switches providers in milliseconds when quotas or rate limits are hit.
- Token compression — RTK + Caveman engine reduces verbose tool outputs (git diffs, logs) by 15–95%, saving an average of 89% on tool-heavy sessions.
- Local first, zero telemetry — Runs on your machine with no cloud middleman; requests go directly from your machine to the provider.
- Compatible with 24+ coding agents — Works with Claude Code, Codex, Cursor, Cline, Copilot, Antigravity, Windsurf, Aider, and any OpenAI-compatible client.
- Quota tracking and sharing — Dashboard shows live used/remaining tokens per provider; Quota-Share feature distributes subscription quotas across team keys.
Who is it for?
- Developers using AI coding assistants — Connect Claude Code, Cursor, or Copilot to a single local endpoint and never be interrupted by rate limits.
- Teams sharing AI subscriptions — Use Quota-Share to divide one Codex Pro or Claude plan across multiple team members fairly.
- Cost-conscious developers — Access free tiers from 90+ providers with no signup fees; 11 providers are free forever (e.g., Kiro AI, Pollinations, OpenCode Zen).
What can you do with OmniRoute?
- Never stop coding: Set up auto-fallback so when one provider runs out of quota, OmniRoute switches to the next in milliseconds, keeping your session uninterrupted.
- Save on API costs: Use free tiers for everyday tasks and reserve paid subscriptions for high-quality models; the system automatically drains subscription tokens before falling back to free.
- Optimize token usage: Enable RTK + Caveman compression to reduce the token cost of verbose tool outputs, making your prompts cheaper and faster.
How does OmniRoute work?
Install OmniRoute via npm install -g omniroute, then run the omniroute command to start the local dashboard at http://localhost:20128. In the dashboard, connect free providers (e.g., Kiro AI, OpenCode Free). Point your coding tool's base URL to http://localhost:20128/v1 with the API key from Dashboard → Endpoints. Set the model to auto for smart routing. OmniRoute then handles fallback and compression automatically.
Pricing
OmniRoute is free and open-source (MIT license) to run on your own machine. Providers themselves may have paid tiers, but OmniRoute aggregates free tokens worth approximately 1.6 billion per month. No credit card is required to start.
Alternatives
- OpenRouter — a paid cloud API that unifies providers behind one hosted endpoint, differing from OmniRoute's self-hosted free approach.
