OpenAI ships GPT-6 Astra

OpenAI released GPT-6 Astra on September 3, its current flagship model. The rollout is deliberately staged — access is being granted across enterprise and developer tiers in phases rather than everything-at-once, a cadence OpenAI has used for recent frontier releases to manage load and monitor for issues before wide availability.

Where it sits in the news cycle

The model carries a Critical cybersecurity rating in OpenAI’s own safety documentation, which puts it in the bracket that requires elevated safeguards before deployment. That rating also made GPT-6 Astra central to the week’s policy story: it’s the model named at the center of the safety-coordination talks with Anthropic and Google, where the labs are negotiating how frontier models get tested and released. OpenAI points to full specs and documentation on its news page for teams planning to build on top of it.

(Original source: OpenAI)