OpenAI launches GPT-6 Astra, its most powerful AI model with advanced cybersecurity capabilities

OpenAI on Thursday unveiled GPT-6 Astra, a new frontier AI model that CEO Sam Altman says represents a major jump in what people can hand off to artificial intelligence. The release arrives with an unusual twist: Astra is so capable at cybersecurity that OpenAI is initially restricting some of its most advanced capabilities.

Announcing the launch on X alongside a video clip, OpenAI summed up Astra’s ambition in one striking claim: “This is GPT-6 Astra. Anything you can do on a computer, Astra can do for you. Fast.”

This is GPT-6 Astra.

Anything you can do on a computer, Astra can do for you. Fast. pic.twitter.com/gDd0IsewJw

— OpenAI (@OpenAI) September 3, 2026

The maker of ChatGPT described Astra as the result of “years of research and big bets,” positioning the model as a step beyond its previous systems in software engineering, computer use, science, professional work and long-running tasks.

“GPT-6 Astra is the most intelligent and aligned model in the world, and sets a new state of the art for computer use, browsing, software engineering, cybersecurity, science, and professional work,” OpenAI said.

Altman told CNBC that Astra represents a “new capability level” and has already changed how he works. He expects the technology to drive “a boom of entrepreneurship, of creativity, of economic growth, of scientific discovery.”

That optimism comes alongside a security challenge OpenAI has rarely faced at this level.

Astra is the first OpenAI model to reach the company’s “Critical” internal cybersecurity threshold, a classification reserved for capabilities that could pose serious risks if misused. Rather than immediately opening every capability to the public, OpenAI is taking a phased approach.

A limited group of companies enrolled in Daybreak, OpenAI’s application-based cybersecurity program, will receive the first access to Astra’s highest-level cyber capabilities.

The broader model is set to reach ChatGPT Plus, Pro, Business and Enterprise customers, along with developers through the OpenAI API and Amazon Web Services, in the coming days.

GPT-6 Astra arrives after OpenAI’s Hugging Face security scare

The timing makes the security restrictions hard to separate from last month’s incident.

Two OpenAI models escaped containment, reached the open internet and breached Hugging Face’s systems. Astra was not involved, but OpenAI temporarily paused parts of its research and training operations after the incident, including work involving Astra.

OpenAI added new safeguards to the model following the breach. The company said Tuesday that those protections “sufficiently minimize the risk of severe harm for release.”

“AI can only benefit people when safety is a core part of it, and so we’re putting more compute and effort towards safety, security, alignment than ever before,” OpenAI President Greg Brockman told reporters Thursday.

Altman said Astra went through a formal review process with the Trump administration before its release.

Beyond cybersecurity, OpenAI says Astra is better at maintaining context, interpreting user intent, respecting task boundaries and completing multi-step workflows. Those improvements point to a broader shift in the AI race. The competition is moving beyond chatbots that answer questions toward systems that can take on chunks of real work with less human supervision.

“There’s still more to do, I think that there’s still lots of improvements to be made, but there is something significant here that I think is qualitatively improved, and that to me is, is what is significant and a real shift in what kind of work people can delegate to AI and how it can empower them,” Brockman said.

That shift could matter just as much commercially as technically.

OpenAI has spent the past year pushing deeper into corporate accounts, competing with Anthropic and Google for enterprise AI spending. CFO Sarah Friar recently told employees that OpenAI’s enterprise unit now generates more revenue than its consumer business, according to CNBC.

The stakes are growing as OpenAI prepares for a possible public-market debut. The company confidentially filed its IPO prospectus with the Securities and Exchange Commission in June. Friar has told employees that OpenAI “will be a public company in 2027,” leaving open the possibility of an earlier listing if the business keeps accelerating.

Astra now becomes part of that story. OpenAI needs to prove that increasingly autonomous AI systems can create economic value without risks that grow as quickly as their capabilities.

GPT-6 Astra may be OpenAI’s strongest demonstration yet of both sides of that equation.

OpenAI