DeepInfra Joins Hugging Face Inference Providers as Routed Serverless Backend

Summary

Hugging Face added DeepInfra as a supported Inference Provider on April 29, 2026, opening another path for developers to run open-weight models through HF's Router with no markup over DeepInfra's direct rates. Initial coverage includes DeepSeek V4 Pro (862B), Kimi-K2.6 (1.1T multimodal), and GLM-5.1 (754B) for chat/text-generation, with text-to-image, text-to-video, and embeddings expected to follow. The integration supports both UI-driven configuration and the existing OpenAI-compatible Python/