Microsoft Launches Foundry Local, Enabling Developers to Integrate On-Device AI Inference

Microsoft has launched Foundry Local to address generative AI applications’ privacy, latency and cost requirements. Developers can embed models in applications and run inference offline on users’ devices, reducing the need to upload data to the cloud or pay ongoing computing costs. The technology could play an important role in enterprise adoption and the broader use of personal AI.

As of July 20, 2026, Microsoft had released the generally available version of Foundry Local, offering cross-platform, on-device AI inference and SDKs for multiple programming languages. It can automatically detect a device’s hardware and select the appropriate runtime to improve efficiency. Microsoft did not announce a separate subscription price, while developers can reduce their reliance on cloud connectivity and lower inference costs.