Source: Hugging Face Inference Providers docs

Hugging Face's serverless router to partner inference providers behind one token and an OpenAI-compatible endpoint, with :fastest/:cheapest/:preferred routing.

Contents
  • What: official product documentation
  • Publisher: Hugging Face
  • Date: accessed 2026-11-14
  • Original: Inference Providers

Summary

Inference Providers is a serverless proxy that sends requests for Hub models to partner inference companies. You use one Hugging Face token and one bill, with no markup on provider rates.

Key claims

  • Partners as of 2026-11 include Baseten, Cerebras, Cohere, DeepInfra, Fal AI, Featherless, Fireworks, Groq, HF Inference, Novita, Nscale, OVHcloud, Public AI, Replicate, Scaleway, Together, WaveSpeedAI and Z.ai. Supported tasks vary by partner.
  • https://router.huggingface.co/v1 is an OpenAI-compatible chat-completions endpoint and handles chat only. The InferenceClient SDKs (Python and JS) cover the other tasks.
  • Routing policy is set with a suffix on the model ID: :fastest (the default, highest tokens/s), :cheapest (lowest price per output token), :preferred (your ordered list), or an explicit provider name such as :groq.
  • With automatic selection, requests fail over to another provider when the primary one is flagged unavailable.
  • Billing is pay-as-you-go with no markup, and PRO and Team/Enterprise accounts get monthly credits. Calls require a fine-grained token with the Inference Providers permission.

Feeds

Hugging Face ecosystem

Suggest changes →