GPT-OSS Model Lands on Windows

Microsoft is integrating OpenAI's gpt-oss-20b model into Windows 11 via its Windows AI Foundry. This platform allows developers to leverage AI features, APIs, and open-source models directly on their computers. The integration, powered by ONNX Runtime, enables local AI inferencing. This offers developers tools to build with open models.

The gpt-oss-20b is a 21 billion parameter model that uses a Mixture-of-Experts (MoE) architecture, activating 3.6 billion parameters per token. It is designed to run efficiently on devices with 16GB of memory, making it suitable for on-device use cases and rapid iteration. The model delivers strong performance on tool use, few-shot function calling and reasoning tasks. It also supports full chain-of-thought (CoT) and structured outputs.

Windows AI Foundry provides developers with a range of tools for AI development, including AI APIs, Windows ML, and the AI Toolkit for Visual Studio Code. It supports customisation of models and deployment across CPUs, GPUs and NPUs. The platform aims to democratise AI development by providing access to ready-to-use open-source models and tools for fine-tuning and optimisation.