Key Info
OpenRouter and Hugging Face announced a partnership that lets chat models on the Hugging Face Hub run through OpenRouter's routing, using the same Hugging Face token and API.
Highlights
- Each request is routed to the provider with the best price, speed, and uptime for that model, with automatic failover.
- The integration covers 160 models from 38 labs.
- Open-weight models grew from 27% to 72% of tokens on OpenRouter in under a year, per the announcement.
- OpenRouter traces can be sent to Hugging Face Storage via Broadcast, so requests sit next to models on the Hub for evals or fine-tuning.
- Open model weights are also accessible directly from OpenRouter.