Microsoft Offers OpenAI’s GPT-OSS-20B for Local Use on Windows

Microsoft has begun distributing OpenAI’s new open-weight language model, gpt-oss-20b, to Windows 11 users through its Windows AI Foundry platform, allowing the model to run locally on PCs equipped with at least 16 GB of graphics memory. The 20-billion-parameter system is tuned for code execution and other tool-oriented tasks, giving developers a way to embed AI assistants in applications even when internet bandwidth is limited. The company said support for macOS and additional devices is in development and that both gpt-oss-20b and the larger gpt-oss-120b are now hosted in Azure AI Foundry. Amazon’s AWS cloud has also listed the models, marking the first time a major Microsoft competitor has gained immediate access to the latest OpenAI technology and underscoring the increasingly open distribution strategy around the partners’ models. In a separate move, Microsoft engineers outlined a phased roadmap to make the Windows UI Library (WinUI) “truly open source.” The team must first separate WinUI from proprietary Windows components, and no target completion date has been set, but the company plans to migrate full development and issue tracking to GitHub once the code is untangled.