DeepInfra Joins HF Inference Providers

DeepInfra has joined Hugging Face Hub as an Inference Provider.

You can now select DeepInfra directly from model pages and the JS/Python SDKs, integrated via huggingface_hub 1.11.2 or later and @huggingface/inference. DeepInfra is a serverless inference platform offering 100+ models, and initially supports conversational and text-generation tasks.

• Open-weight LLMs such as DeepSeek V4, Kimi-K2.6, and GLM-5.1 are available.
• text-to-image, text-to-video, and embeddings will be added later.
• It also connects with Agent Harnesses such as Pi, OpenCode, Hermes Agents, and OpenClaw.

In account settings, you can register a custom API key per provider, and using an HF token will automatically route requests to DeepInfra. You can also set provider priority, which the model page widget and code snippets will follow.

Billing is handled by the provider for direct requests, and by Hugging Face at the provider's exact rate for routed requests. PRO users receive $2 in inference credits each month, and logged-in free users also receive a small amount of free usage.