OpenAI unveiled its new AI model, GPT-6 Astra, on September 3, 2026. Astra is a model tailored for complex reasoning, coding, web browsing, computer usage, and document creation. OpenAI announced that it will first provide it to the enterprise Trusted Access Program and expand it to APIs and ChatGPT Plus, Pro, Business, and Enterprise plans within a few days.
Astra has been classified as the first model to reach a Critical level of cybersecurity capability in OpenAI's readiness assessment framework. OpenAI explained that, provided it is provided with the appropriate tools and access permissions, this model can identify unknown security vulnerabilities and create ways to exploit them. Accordingly, advanced cybersecurity features will initially be available only to a limited number of testers.
OpenAI announced that it delayed some development and release schedules prior to the launch of Astra and strengthened security measures. These include denying malicious cyber requests, detecting misuse, and monitoring unauthorized behavior. The company explained that while Astra was not directly involved in the HuggingFace incident, it incorporated lessons learned from the event into its security design.
See more and sources
https://openai.com/index/gpt-6-astra/
