New models released by OpenAI, Anthropic, and Google

• OpenAI, Anthropic, and Google have released new AI models
• The models from OpenAI received mixed reactions from users
• Anthropic and Google also presented their updates

On September 6, 2025, three companies, OpenAI, Anthropic, and Google, announced the release of new artificial intelligence models, which generated significant interest in the technological community.

OpenAI introduced two new models — gpt-oss-120B and gpt-oss-20B, which are the first releases with open weights since 2019. These models, although licensed under Apache 2.0, are not fully open, as the data for training remains restricted. The gpt-oss-120B model can operate on powerful hardware, while gpt-oss-20B is designed for use in simpler environments, such as notebooks.

Users have already begun testing the new models, and reactions have been mixed. In fact, they demonstrate high performance in mathematics and programming, while others indicate limitations in creative writing and overall quality.

Anthropic also released an updated model, Claude Opus 4.1, which includes enhanced capabilities in real-time coding and reasoning. This model continues to demonstrate the company's growth, which has increased its revenue fivefold in the last seven months.

Google introduced Genie 3, a new model capable of creating interactive gaming worlds in real time. This model, according to the developers, has improved memory and visual accuracy.

All of these new models are already available for public use, and their launch has sparked significant interest among professionals in the technology sector.

Tags: Technology/AI