DeepInfra Now Supported as Hugging Face Inference Provider
Patch your Hugging Face SDK to >=1.11.2, update model identifiers to include :deepinfra, configure API keys or routing, and test inference with example code.
Patch your Hugging Face SDK to >=1.11.2, update model identifiers to include :deepinfra, configure API keys or routing, and test inference with example code.
Summary
DeepInfra has been added as a supported inference provider on the Hugging Face Hub as of 29 April 2026, expanding the range of serverless inference options available directly on model pages. The integration brings over 100 models, including conversational and text‑generation models such as DeepSeek V4, Kimi‑K2.6, and GLM‑5.1, into the Hugging Face ecosystem. It is available through the Hugging Face SDKs, requiring huggingface_hub >=1.11.2 for Python and @huggingface/inference >=1.11.2 for JavaScript. Two modes are supported: a custom key mode that routes calls directly to DeepInfra using your own API key, and a routed mode that uses the HF token and bills through HF. Billing follows the provider’s rates for direct requests and HF’s standard rates for routed requests, with no markup. PRO users receive $2 worth of inference credits per month, and free users have a small quota. Example code snippets for Python and JavaScript demonstrate how to call a DeepSeek V4 Pro model via the DeepInfra provider.
The announcement includes a dedicated documentation page and a full list of supported models. DeepInfra’s pricing is described as the most cost‑effective per token in the industry, and the integration is positioned to simplify the developer experience by eliminating glue code for multiple providers.
Key changes
- DeepInfra added as inference provider on Hugging Face Hub
- Supports conversational and text‑generation tasks with models like DeepSeek V4, Kimi‑K2.6, GLM‑5.1
- Over 100 models available
- Integration into HF SDKs (huggingface_hub >=1.11.2, @huggingface/inference JS)
- Two modes: custom key (direct to provider) or routed via HF
- Billing follows provider or HF rates with no markup
- PRO users receive $2 inference credits per month
- Example code snippets for Python and JS using base_url https://router.huggingface.co/v1