Hugging Face moves Inference Providers routing to OpenRouter Hugging Face has moved its Inference Providers routing for language models to OpenRouter, sending every chat-model request through OpenRouter's router while keeping the existing Hugging Face token and API unchanged. OpenRouter, which serves more than 800 trillion tokens a month to over 10 million users across 600+ models from 80+ providers, says its rebuilt routing cut capacity errors by 46% while improving time to first token by 16.9% and throughput by 16.3%. Pricing is unchanged with no markup, PRO users get $2 in monthly credits and Team and Enterprise organizations get $2 per seat, but custom provider API keys no longer work for language models, though they still work for image and video providers. Hugging Face has switched its Inference Providers service from its custom router to OpenRouter, sending every language model request through OpenRouter in a new partnership designed to make open-weight models easy to run. Developers can now call any open-weight model on the Hugging Face Hub that a provider serves with their existing Hugging Face token and the same API, while OpenRouter picks the best provider for each request based on live price, speed, and reliability. Existing code stays the same. The router supports OpenAI's Chat Completions and Responses APIs, as well as Anthropic's Messages API, for every chat model, regardless of the provider behind it, and Hugging Face offers setup guides for Claude Code, Codex, OpenCode, Pi, and Hermes Agent. When a provider slows down or fails, traffic moves away from it until it recovers, and a failed attempt is retried on the next best provider inside the same call. Previously, Inference Providers picked a single provider and returned its error. Users can still pick a routing policy or a specific provider by adding a suffix such as :fastest , :cheapest or a provider name to the model id. Routing is also cache-aware, keeping a conversation on the provider that holds its prompt cache, with an optional X-HF-Session-Id header to make that explicit. Pricing does not change. Users pay the provider's price on their Hugging Face bill with no markup; PRO users get $2 in monthly credits, and Team and Enterprise organizations get $2 per seat. Model pages gain new Providers and Metrics tabs covering price, speed, uptime, throughput, time to first token and context length, usage overviews, and a breakdown of spend by provider, and both users and organizations can turn off providers they do not want. One breaking change: custom provider API keys no longer work for language models, though they still work for image and video providers. Image, video, and audio generation continue to run directly through Hugging Face Inference Providers without OpenRouter. OpenRouter serves more than 800 trillion tokens a month to over 10 million users across 600+ models from 80+ providers, and says its rebuilt routing cut capacity errors by 46% while improving time to first token by 16.9% and throughput by 16.3%. Demand is shifting toward these models, as open-weight models rose from 26.8% of tokens served on OpenRouter in October 2025 to 71.7% at the end of September 2026, and cost about four times less on average than closed models. Hugging Face calls this the first step of the partnership, with OpenRouter account linking and more tasks routed through OpenRouter coming next. Sources and related context - OpenRouter Now Powers Inference on Hugging Face https://openrouter.ai/blog/announcements/hugging-face-inference-providers/?ref=testingcatalog.com : Supports: OpenRouter's own announcement confirms that chat models on Hugging Face Inference Providers now run through its routing, with failover, model id suffixes and the new Providers tab. - How OpenRouter rebuilt provider routing for better, faster inference https://openrouter.ai/blog/announcements/how-we-rebuilt-provider-routing/?ref=testingcatalog.com : Supports: OpenRouter's October routing write-up details the 46% capacity error, 16.9% time to first token and 16.3% throughput figures and the switchback experiment behind them. - Routing and Providers in the Hugging Face Inference Providers docs https://huggingface.co/docs/inference-providers/routing?ref=testingcatalog.com : Supports: Hugging Face's documentation describes the default routing, the fastest, cheapest and provider-name suffixes, provider opt-outs and failover covered in the article.