A unified LLM API gateway. One OpenAI-compatible endpoint, one key, 40+ supported models — GPT, Claude, Gemini, DeepSeek, Mistral, Llama and more — at up to 90% off official prices, with smart routing, automatic provider fallback, and a cost trace for every request.
Calling an LLM directly is a black box. Through a gateway, every call has a ledger (what it cost), a trace (which model and route served it), and control (per-key budgets and rate limits).
| Protocol | Base URL | Use with |
|---|---|---|
| OpenAI-compatible | https://api.router.one/v1 |
OpenAI SDKs, LangChain, Vercel AI SDK, Codex CLI |
| Anthropic-compatible | https://api.router.one |
Claude Code, Anthropic SDKs |
Native Claude Code and Codex CLI support — no wrapper, no proxy, two environment variables. Reachable worldwide, including mainland China. No prompt or completion storage: the gateway logs metadata only.
- claude-code-codex-setup — Point Claude Code & Codex CLI at the gateway in 2 minutes, with config templates and troubleshooting
- router-one-examples — Runnable examples: curl, Python, Node, LangChain, Vercel AI SDK, Anthropic SDK
- awesome-llm-gateway — A curated list of LLM gateways, proxies, routers, and observability tooling
🌐 router.one · 📖 API docs · 🗂 Model catalog & rates · 🧭 Error codes reference · 🐦 @router_one
