OpenAI has unveiled two new open-weight models, gpt-oss-120b and gpt-oss-20b, marking a significant shift in the company’s strategy and in the wider open-source AI ecosystem.
The models, released on Tuesday, represent OpenAI’s first official open-weight offerings, allowing developers, researchers, and even hobbyists to run advanced AI models locally, with no internet connection required.
The gpt-oss-120b, boasting 120 billion parameters, delivers reasoning capabilities on par with OpenAI’s o4-mini model. Despite its scale, it is optimized to run on a single 80GB GPU, making it viable for academic labs, startups, and well-resourced developers.
Meanwhile, the lighter gpt-oss-20b matches the performance of o3-mini and is capable of running on edge devices with as little as 16GB of memory, opening the door to powerful offline AI applications on personal laptops and compact systems.
For non-technical users, this means a powerful shift: AI that previously required cloud-based supercomputers can now run directly on local machines, boosting privacy, security, and accessibility.
OpenAI CEO Sam Altman described the release as a milestone for global innovation and digital sovereignty.
“We’re excited to make this model, the result of billions of dollars of research, available to the world to get AI into the hands of the most people possible,” Altman said. “As part of this, we are quite hopeful that this release will enable new kinds of research and the creation of new kinds of products.”
He added that OpenAI was “excited for the world to be building on an open AI stack created in the United States, based on democratic values, available for free to all and for wide benefit.”
