OpenAI has launched Astra, which the company describes as its most capable model to date, while placing its most advanced cybersecurity capabilities behind a staged access plan. Astra has reached OpenAI’s "critical cybersecurity capability threshold" under its Preparedness Framework, according to the company. Initial access to advanced cyber work is limited to a small group of testers through Daybreak Blue, with expansion toward defensive use planned as OpenAI’s safety controls mature.
The announcement shifts the story beyond an early availability teaser. In its official Path to Astra announcement, dated September 1, 2026, OpenAI sets out both the model’s capability classification and the restrictions governing its release. For businesses, the practical takeaway is straightforward: Astra is a confirmed OpenAI release, but its most sensitive capabilities are not broadly available at launch.
What OpenAI has confirmed about Astra
Astra is significant because OpenAI has formally classified it at the company’s critical cybersecurity capability threshold. That designation is tied to the Preparedness Framework, OpenAI’s approach to evaluating and managing risks from increasingly capable models. The company is therefore pairing the release with limited access and additional safeguards rather than treating all capabilities as a standard, immediately available product feature.
