GPT-6 Astra Is Rolling Out in Phases, and Cybersecurity-Programme Customers Go First

OpenAI's newest model reaches paid ChatGPT tiers, the API and AWS over the coming days — but access starts with companies in its application-based cybersecurity programme, because of what the model can do in that area.

OpenAI began rolling out GPT-6 Astra on 3 September, and the order of the rollout is the part with information in it: the first users are companies participating in OpenAI's application-based cybersecurity programme, not the general paid base.

From there the model becomes available to ChatGPT Plus, Pro, Business and Enterprise users, through the OpenAI API, and on Amazon Web Services.

Why the gating exists

OpenAI has said the model's cyber capabilities are advanced enough to warrant additional internal safeguards, and it is releasing a version with added cybersecurity guardrails to paid ChatGPT users. Both versions carry protections intended to keep users away from the most capable end of that behaviour.

A staged rollout that starts with vetted security customers, rather than with the largest paying tier, is a decision about who is trusted with the capability first.

What the company said about it

CEO Sam Altman told CNBC that Astra was a "new capability level" and had changed his own workflows. OpenAI described the model as the product of "years of research and big bets."

President Greg Brockman called it a generational leap and said people may look back on this model as the point at which AGI arrived. That is a claim about the future made by the company selling the model, and no benchmark released alongside it settles it.

Sources: CNBC, 3 September 2026 · Venture Capital channel