Hugging Face said on November 16, 2026 that it is switching its Inference Providers service for language models from its own custom router to OpenRouter's. In a blog post titled "Hugging Face and OpenRouter join forces to make open-weight models easy to run," the company said requests to language models now go through OpenRouter, which picks the best provider for each call based on live price, speed and reliability and retries on another provider if the first one fails. It said developers' code stays the same.

Hugging Face says every open-weight model on the Hub that a provider serves can now be called with a Hugging Face token through one API. OpenRouter routes across more than 80 providers, and the router supports OpenAI's Chat Completions and Responses APIs and Anthropic's Messages API for every chat model. The post includes setup guides for Claude Code, Codex, OpenCode, Pi and Hermes Agent. Model IDs accept suffixes such as :fastest, :cheapest or a provider name to override the default policy, and routing is cache-aware, keeping a conversation on the provider that holds its prompt cache, with an X-HF-Session-Id header available to make this explicit.

The post cites performance figures from OpenRouter. Hugging Face says OpenRouter's rebuilt routing cut capacity errors by 46% and improved time to first token by 16.9% and throughput by 16.3% in OpenRouter's own measurements. It also says OpenRouter serves more than 800 trillion tokens a month to over 10 million users across 600 or more models and that, on OpenRouter, open-weight models rose from 26.8% of tokens served in October 2025 to 71.7% at the end of September 2026. It reports that open-weight models cost about four times less on average and that the best open-weight model, Kimi K3.1 Pro, scores 46 on the Artificial Analysis Intelligence Index against 65 for Claude Fable 5.2 at about one-seventh of the cost per task. Those statistics are attributed by Hugging Face to OpenRouter and Artificial Analysis and were not independently checked.

Pricing does not change. Users pay the provider's price on their Hugging Face bill with no markup, and PRO users receive $2 of monthly credits, with Team and Enterprise organizations receiving $2 per seat. Image, video and audio generation continue to run directly through Inference Providers and do not go through OpenRouter.

The change has one breaking effect. Because language-model requests are now billed to the Hugging Face account or organization through OpenRouter, users can no longer bring their own API key for a provider with these models. They can still select a provider with the model:provider suffix or disable providers in settings, and custom keys for image and video providers keep working.

Hugging Face describes this as the first step of the partnership, with plans to let users connect an OpenRouter account to Hugging Face and to route more tasks through OpenRouter.

References