OpenAI has launched its flagship AI model, GPT-6 Astra, which outperforms all competitors in benchmark tests. This groundbreaking model is capable of creating three-dimensional urban scenes in Unity, generating animated models in FreeCAD and Blender, and even autonomously filling out tax returns.
Astra excels particularly in coding tasks and identifying vulnerabilities in software. In the DeepSWE v1.1 test, which evaluates a model's ability to independently solve real-world software development challenges, GPT-6 Astra achieved a remarkable score of 74.1%. This surpasses Claude Fable 5.1 by 6.7% and exceeds Gemini 3.8 Flash by 0.1%.
Greg Brockman, President of OpenAI, believes that with GPT-6 Astra, the company has reached a level of artificial general intelligence (AGI), although he emphasizes that users should determine if the model meets this definition. Notably, GPT-6 Astra is the first model trained using over 100,000 GPUs at the Stargate data center in Texas.
As of today, GPT-6 Astra will be accessible to a limited number of organizations, with broader availability expected in the coming days for users of ChatGPT Plus, Pro, Business, and Enterprise. This advancement sets a new standard in the AI market, potentially reshaping the competitive landscape and prompting rivals to accelerate their innovations.
Informational material. 18+.
