Microsoft Foundry Local: On-Device AI Runtime With Zero Cloud Dependency — 2,059 Stars

GitHub·medium signal

Microsoft's Foundry Local enables multimodal AI models (text, image, audio) to run on local NVIDIA GPU hardware with zero cloud connectivity. APIs mirror the cloud surface — Responses API, function calling, agent services — same code, different runtime. Targets government, defense, finance, and healthcare with strict data sovereignty. February 2026 update added support for large multimodal models and the NVIDIA Vera Rubin platform.