Hugging Face routes language model requests through OpenRouter
Original title: Hugging Face now routes model requests through OpenRouter for the best price or speed
Bring-your-own keys for language models no longer work, while image and video keys remain.
Hugging Face's Inference Providers service now sends language-model requests through OpenRouter, which picks a provider by price, speed, and reliability and fails over during outages. Existing code keeps working, and suffixes or the X-HF-Session-Id header can steer provider choice. Bring-your-own API keys for language models are going away, while image, video, and audio generation still runs directly. The article also reports Nvidia's planned purchase of Hugging Face and Stripe's agreement to acquire OpenRouter, neither of which has closed.
Unconfirmed: Routing gains are OpenRouter's own measurements, and the Nvidia and Stripe acquisitions are reported and not closed.
- The :fastest suffix picks highest throughput and :cheapest picks the lowest price.
- OpenRouter says capacity errors fell 46 percent with the new routing.
- Hugging Face says open models rose from 26.8 percent to 71.7 percent of top-model tokens.
- Reports say Nvidia is buying Hugging Face for $12.9 billion, under DOJ review.
- Stripe agreed in August to acquire OpenRouter for more than $7 billion.
- Products
- Inference ProvidersOpenRouter
- Organizations
- Hugging FaceNvidiaStripeU.S. Department of Justice
- AI models
- Kimi K3.1 ProClaude Fable 5.2
From the original article
3 min read at the-decoder.com
Hugging Face now routes language model requests through OpenRouter, which automatically selects the cheapest or fastest provider. Bring-your-own API keys for language models are going away. According to OpenRouter, the new routing cut capacity errors by 46 percent.
By Jonathan Kemper · the-decoder.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.