Baseten on Hugging Face Inference Providers 🔥

Baseten has been announced as a supported Inference Provider on the Hugging Face Hub. According to the announcement, Baseten joins a growing ecosystem that enhances serverless inference directly on model pages and integrates with client software development kits for both JavaScript and Python. As an AI infrastructure platform covering serverless AI and training, Baseten is launching its initial integration with support for conversational and text-generation tasks. This enables user access to popular open-weight large language models including Kimi K3, DeepSeek V4 Flash, and GLM-5.2, with support for additional tasks scheduled to roll out soon. The integration allows developers to use Baseten-hosted models within Hugging Face SDKs, such as hugging_hub for Python and @huggingface/inference for JavaScript, by using a Hugging Face token for authentication. Furthermore, Hugging Face Inference Providers are integrated into agent harnesses like Pi, OpenCode, Hermes Agents, and OpenClaw, enabling direct use without extra code. Regarding pricing, users authenticating via the Hugging Face Hub pay standard provider API rates without additional markups, while direct requests using provider keys are billed directly by the corresponding provider. Additionally, Hugging Face PRO users receive monthly inference credits that can be used across providers, and free users retain access to free inference with a small quota.