Hugging Face routes language model inference through OpenRouter
Hugging Face now uses OpenRouter to select the best provider for each request, keeping the same API and token.
Hugging Face has replaced its custom inference router with OpenRouter for all language model calls.
The new system sends each request to the provider with the optimal mix of price, speed and reliability, based on live measurements from more than 80 providers.
If a provider slows down or fails, traffic is automatically shifted to the next best provider and the failed attempt is retried within the same call.
Developers keep the same Hugging Face token and API; they can still force a routing policy or specific provider by adding a suffix such as :fastest or :cheapest to the model ID.
Image, video and audio generation remain outside OpenRouter and continue to run directly through Hugging Face Inference Providers.
What it means for you
The change improves reliability and cost efficiency for users running open‑weight models, while preserving existing code.
Sources
- Hugging Face moves Inference Providers routing to OpenRouter, TestingCatalog, 17 Nov 2026
- Hugging Face and OpenRouter join forces to make open-weight models easy to run, Hugging Face (official), 16 Nov 2026
Written automatically from the sources above, then checked by our code: every number and name must appear in them, and no sentence may copy a source. Spotted a mistake? Report it.
Which AI can you use for free today?
Every free AI model, ranked by blind human votes, with the memory and speed it needs on your own GPU.