Baseten is now a supported Inference Provider on the Hugging Face Hub, enhancing serverless inference capabilities directly on model pages. This integration allows developers to access Baseten's catalog of frontier models through Hugging Face's client SDKs and UI.

  • Initial support covers conversational and text-generation tasks for open-weight LLMs including Kimi K3, DeepSeek V4 Flash, and GLM-5.2.
  • Users can set custom API keys or use Hugging Face routed requests billed to their HF account.
  • Baseten is integrated into client SDKs (huggingface_hub >= 1.26.1 for Python and @huggingface/inference for JavaScript).
  • The provider is supported by agent harnesses such as Pi, OpenCode, Hermes Agents, and OpenClaw.

This integration enables users to plug Baseten-hosted models into their applications with minimal setup and access popular open-weight LLMs directly through the Hugging Face ecosystem.