OpenAI launched a brand new free and open GPT model yesterday that may run on a PC, and now Microsoft is making that straightforward to do for Home windows customers. The light-weight gpt-oss-20b mannequin is now out there on Home windows AI Foundry, and will probably be coming quickly to macOS, too.
You’ll want a PC or laptop computer with at the very least 16GB of VRAM, so that you’ll want one of many prime GPUs from Nvidia or the number of Radeon GPUs with ample VRAM. The gpt-oss-20b mannequin is optimized for code execution and power use, and Microsoft says it’s “good for constructing autonomous assistants or embedding AI into real-world workflows, even in bandwidth-constrained environments.”
Microsoft has pre-optimized gpt-oss-20b for native inference, and it hints that help for extra gadgets is coming quickly. That would imply we see a extra optimized model for Copilot Plus PCs in some unspecified time in the future sooner or later, very similar to how Microsoft has been adding a variety of local AI models to Home windows just lately.
Microsoft’s speedy addition of OpenAI’s income model mannequin to the Home windows AI Foundry comes as Amazon was equally quick to undertake the brand new open-weight GPT-OSS fashions for its cloud providers. It’s the primary time you’ll be able to run an OpenAI mannequin regionally on Home windows, however it’s additionally the primary time Microsoft’s greatest cloud competitor has had entry to the income model OpenAI fashions — including one other dynamic to the complicated OpenAI and Microsoft partnership.