OpenAI has unveiled a new, free, and open GPT model, designed to operate seamlessly on personal computers. In a significant move, Microsoft is facilitating this integration for Windows users through its local foundry initiatives. The lightweight gpt-oss-20b model is now accessible via the Windows AI Foundry, with plans to extend its availability to macOS shortly.
To utilize this model, users will require a PC or laptop equipped with a minimum of 16GB of VRAM, necessitating high-performance GPUs from Nvidia or select Radeon models. The gpt-oss-20b model has been specifically optimized for code execution and tool utilization, making it an ideal choice for developing autonomous assistants or embedding AI into practical workflows, particularly in environments where bandwidth may be limited.
Microsoft has taken steps to pre-optimize the gpt-oss-20b for local inference, hinting at the potential for broader device support in the near future. This could pave the way for a more refined version tailored for Copilot Plus PCs, reflecting Microsoft’s ongoing commitment to enhancing local AI capabilities within Windows.
The swift incorporation of OpenAI’s latest model into the Windows AI Foundry coincides with Amazon’s prompt adoption of the new open-weight GPT-OSS models for its cloud services. This marks a pivotal moment, as it is the first instance where an OpenAI model can be run locally on Windows. Additionally, it represents a new chapter in the evolving relationship between Microsoft and its primary cloud competitor, Amazon, as both entities gain access to the latest OpenAI models.
