2025-08-05T18:01:37z
- Openai CEO Sam Altman has announced the GPT-Oss. GPT-Oss is back with his company releasing an AI language model for Open Weights.
- The company launched two new “open” models on Tuesday: the GPT-OSS-120B and GPT-OSS-20B.
- Altman previously said Openai would release a “very powerful open source model.”
Openai's AI model is more open. At least a few Those are them.
Openai CEO Sam Altman announced the GPT-Oss on Tuesday. This is a family of “open” language models with “open weights” that the CEO said can be operated locally on “high-end laptops” and smartphones.
AI models with “Open Weights” are “open weights” models to allow AI models to be run, inspected or tweaked locally, as they expose fully trained parameter weights.
“We believe this is the best and easiest open model in the world,” Openai CEO Sam Altman wrote in X.
There are two different models: GPT-OSS-120B and GPT-OSS-20B. Smaller models are designed to run on “most desktops and laptops,” while larger models are directed towards high-end equipment.
Altman said the GPT-Oss has “strong real-world performance comparable to the O4-Mini.”
Just before Openai's announcement, rival humanity revealed Claude Opus 4.1.
Tuesday's announcement wasn't the long-difficult ChatGPT-5 that could arrive soon this week. Instead, the new model is OpenAI's first open-weight language model since the release of GPT-2 in 2019.
“As part of this, we are very excited that this release will allow for new kinds of research and the creation of new kinds of products,” writes Altman. “We expect a meaningful rise in the speed of innovation in our field, and we expect more people to do more important jobs than before.”
Altman previously showed Openai would return to releasing at least some open models.
Meta has been revealing the open weights of its Llama AI models for a long time.
