Hugging Face Models Deployable on Foundry Managed Compute

Microsoft announced at Build 2026 that Hugging Face models are now available on Foundry Managed Compute — a curated catalog of open-weight models refreshed weekly and deployable in one click. The weights are pre-staged in Azure, runtimes are built and scanned by Microsoft, and all models ship with enterprise security, governance, observability, and billing features.

Foundry Managed Compute is a managed GPU platform-as-a-service for open-source and custom models. Users deploy by specifying parameter count, context length, and optimization preferences (latency or throughput), while Foundry handles GPU topology automatically. Microsoft manages container updates, runtime upgrades, and security patches for supported runtimes including vLLM, SGLang, TensorRT-LLM, NIM, TEI, and llama.cpp.

The deployment option shares a single endpoint, SDKs (Python, C#, JavaScript, Java), authentication, and observability tools with Foundry's pay-per-token and provisioned throughput options. Additional platform features include content safety filters, task-adherence guardrails, an AI Red Teaming Agent for adversarial testing, unified RBAC, private networking, and Azure Policy integration.