What changed
Hugging Face announced that its Inference Providers service now routes language model requests through OpenRouter. The routing system automatically selects a provider based on price, speed, and reliability, with the ability to fall back to another provider during outages. Developers can influence routing by using suffixes like ":fastest" or ":cheapest" or by specifying a provider directly. Existing code remains functional without changes.
What this means in practice
Developers no longer need separate API keys for language model providers, as Hugging Face handles billing directly without markup. Pro users receive a small monthly credit, while team and enterprise organizations get credits per seat. Open-weight models now account for over 71 percent of tokens among OpenRouter's most-used models, as they cost roughly a quarter of what proprietary options charge on average. New model pages include tabs showing available providers and their performance metrics, with the option to disable specific providers in settings.
Limitations and availability
Bring-your-own API keys for language model providers are no longer supported, though they remain available for image and video providers. Image, video, and audio generation continues to run through Inference Providers. The routing system does not affect these tasks. The change does not specify any hardware requirements or license conditions. The supplied announcement does not specify the exact model versions or benchmark conditions used for performance improvements.
