Hugging Face Inference
Hosted open models
Last verified: 1 October 2026
Hugging Face Inference Providers give access to hundreds of models through a single API, routed to partner providers (Groq, Together, Replicate, Fireworks, Cohere, etc.). They are integrated into the Python and JavaScript SDKs and offer an OpenAI-compatible chat endpoint, with no markup on provider rates.
Services
- Text, image and video generation
- Embeddings and classic tasks (NER, summarization, speech recognition)
- Provider selection: fastest, cheapest or preferred
- Inference Playground
Access requirements
A Hugging Face token is required; a free tier exists, with additional credits for PRO, Team and Enterprise accounts.
Sandbox: Yes
Sources
- Hugging Face Inference (new tab), verified on 1 October 2026Official site consulted