OpenRouter is a router for language models. You send your request once, and it picks which provider answers it, based on price, speed and reliability measured right then. If one fails, the try is repeated on the next. Your code stays the same.
On November 16, 2026, Hugging Face switched its Inference Providers from its own router to OpenRouter's. Requests to language models now go through it, and the code of everyone using it stays the same.
Why one router is enough for everything
The point is simple. The same model can be served by several providers, at different prices and different response times. The router measures who is doing well right now and sends the request there. In OpenRouter's measurements, the new routing cut capacity errors by 46% and improved time to first token by 16.9%.
One address, many providers. The choice is made on every request, not once and for all.
The scale: by OpenRouter's numbers, the service handles more than 800 trillion tokens a month for over 10 million users, across 600+ models from 80+ providers.
Open-weight models went from 26.8% of tokens in October 2025 to 71.7% at the end of September 2026, after peaking at 77.5% in July. Their weekly volume grew more than 60-fold, against almost tenfold for closed models. The most used model there is an open one, DeepSeek V4.1 Flash 1017, with 59.2 trillion tokens.
Price is the reason. By OpenRouter's numbers, open-weight models cost about four times less on average: $0.28 per million input tokens and $1.62 per million output tokens, against $1.27 and $6.36 for closed models. The best open model still trails on the hardest tests: Kimi K3.1 Pro scores 46 on the Artificial Analysis index against 65 for Claude Fable 5.2, at about a seventh of the cost per task.
The numbers are OpenRouter's, not Hugging Face's. The open-weight share is a share of traffic through one router, not of the whole world.
If you use Hugging Face, there is one breaking change: your own provider key no longer works with language models. Requests go through OpenRouter and are billed to your Hugging Face account. Prices don't change, with no markup. You can pick a provider with a suffix on the model id (:fastest, :cheapest) or turn a provider off in your settings. Keys for image and video keep working.
Primary sources
the Hugging Face announcementThe visual is generated code-art, no third-party images.