Microsoft is making OpenAI’s new free and open GPT mannequin, gpt-oss-20b, accessible to Home windows 11 customers by way of Windows AI Foundry, the tech big’s platform that lets customers faucet AI options, APIs, and fashionable open-source fashions on their computer systems.
The corporate mentioned in a blog post that gpt‑oss-20b is “tool-savvy and light-weight,” including: “Optimized for agentic duties like code execution and gear use, it runs effectively on a variety of Home windows {hardware}, with assist for extra units coming quickly. It’s excellent for constructing autonomous assistants or embedding AI into real-world workflows, even in bandwidth-constrained environments.”
Launched on Tuesday, OpenAI’s gpt-oss-20b can run on shopper PCs and laptops with not less than 16GB of VRAM, which a contemporary GPU from Nvidia or Radeon would have. OpenAI mentioned the mannequin was educated utilizing high-compute reinforcement learning, which helps it excel at powering AI brokers and name instruments, reminiscent of internet search or Python code execution, as a part of its chain-of-thought course of.
Nevertheless, the mannequin is text-only, that means it won’t be able to course of or generate photographs and audio like the corporate’s different fashions.
Additionally, the mannequin has a behavior of hallucinating loads: gpt-oss-20b hallucinated in response to 53% of questions on PersonQA, OpenAI’s in-house benchmark for measuring the accuracy of a mannequin’s information about individuals.
Microsoft says it should convey the mannequin to macOS quickly, in addition to different units, although the corporate didn’t identify any extra. Microsoft can be making each gpt-oss-20b and older mannequin gpt-oss-120b accessible by way of its hosted Azure AI Foundry platform.
The 2 new fashions are additionally available on Amazon’s AWS.
Techcrunch occasion
San Francisco
|
October 27-29, 2025