RouterBench

•  the catalog  •

130+ models.
One API.

Every major model across 25+ providers — plus RouterBench's own models and your custom, fine-tuned or self-hosted endpoints. All routable, all scored, all behind one endpoint.

•  providers  •

Every major provider. One bill.

OpenAIAnthropicGoogleGroqMoonshotZ.aiSarvamMistralxAIMeta LlamaQwenPerplexityCerebrasFireworksTogetherNovitaAINebiusRouterBench (our models)Your endpoints

•  what's routable  •

If it speaks the protocol, it's a candidate.

01 — frontier

Hosted frontier models

GPT, Claude, Gemini, Grok, Llama, Qwen, Kimi and more — with multiple hosts per model and automatic provider selection.

03 — yours

Custom & fine-tuned

Register any OpenAI/Anthropic-compatible endpoint — Ollama, vLLM, LM Studio, TGI — as a first-class provider.

The catalog is live.

Filter by task, price, context and score in the console — or just send routerbench/auto and let the router pick.