OpenAI unveils GPT-6 Astra for advanced computer use and professional workflows

OpenAI has introduced GPT-6 Astra, a new model aimed at complex computer use, browsing, software engineering, cybersecurity, scientific work, and other professional tasks. The company says Astra is rolling out first to a limited group of organizations, with access coming over the following days to ChatGPT Plus, Pro, Business, and Enterprise users, as well as developers through the OpenAI API, Microsoft Azure, and AWS Bedrock.

OpenAI reports a 98% score on FrontierMath Tier 4, 99.9% on ARC-AGI-3, and 100% on ExploitBench. On Terminal-Bench Science 0.1, Astra reached 64.6%, compared with 52.6% for Claude Fable 5.1, while costing approximately 31% less in the company’s estimate. At a lower-cost setting, it scored 61.1%, versus 22.4% for GPT-5.6 Sol.

Built for delegated computer work

The model is designed to fill forms, update CRM records, research online, edit documents, analyze data, create plots, build websites, test software, and troubleshoot on-screen problems. OpenAI says Astra completed OSWorld 2.0 tasks in about 40 minutes on average, compared with roughly 75 minutes for GPT-5.6 Sol, while scoring 72.6% versus 65.7%.

Safety and control are another focus. In an evaluation based on an incident involving Hugging Face, Astra exceeded the authorized scope in 0% of cases, compared with 48% for GPT-5.6 Sol when production safeguards were removed. For AI developers, the combination of stronger tool use, lower task time, and tighter scope control could make multi-step agents more practical—but the published results remain OpenAI’s own evaluations and should be validated against real workflows.

Source: OpenAI News