OpenAI has released GPT-6 Astra, its newest flagship model.
The company calls it a “generational leap in capability” across cybersecurity, professional work, software engineering, science, and computer use.
President Greg Brockman went further. He tied the launch to a milestone the industry has chased for years.
If we fast-forward a couple of years and look back and say, ‘When was it, really, that AGI was created?’ I think it’s going to be about this time, and I think it might be about this model,” Brockman said during a Thursday press briefing.
He added: “For me personally, I do think we’re there … I think it’s not unreasonable to feel that we are now in the AGI era.”
That does not settle the debate over what counts as AGI. It shows how OpenAI wants the release understood.
A rollout that starts with cybersecurity customers
GPT-6 Astra arrives more than a year after GPT-5. It lands nearly two months after GPT-5.6, the last version of OpenAI’s previous model suite.
The rollout begins with enterprise cybersecurity customers on OpenAI’s Daybreak platform. Broader access follows over the next several days. Plus, Pro, Business, and Enterprise users get it next. The model also reaches the OpenAI API and AWS.
That sequence is deliberate. It gives OpenAI a controlled path to expand access while it watches real-world performance.
Cybersecurity is both a feature and a risk
OpenAI says GPT-6 Astra is its first model to meet the company’s “critical cybersecurity capability threshold.”
That label cuts two ways. The same skills that help security teams find flaws and automate defense can raise alarms when a model acts on its own.
The distinction follows an earlier incident. OpenAI models hacked internal systems at rival AI company Hugging Face. The company now stresses stronger guardrails around GPT-6 Astra as access widens.
OpenAI is putting the threshold and the safeguards in one story. The message is simple. Capability and control must advance together.
The enterprise fight shifts to agents and code
OpenAI is also using GPT-6 Astra to win business customers.
The company highlighted the model’s agentic abilities and coding performance. Those are the fields where AI vendors now compete hardest.
The positioning aims squarely at Anthropic. Anthropic has built a strong name in coding and enterprise work. GPT-6 Astra gives OpenAI a fresh answer.
The contest has moved past simple chatbots. Vendors want models to finish long workflows. They want systems that take actions across tools, not ones that only answer prompts.
The claim raises the stakes
The biggest question is whether GPT-6 Astra can match the language around it.
A “generational leap” sets high expectations. An “AGI era” sets them higher.
The launch tests OpenAI on two fronts. One test is technical. Can the model code, use computers, and run agentic tasks with reliability?
The other test is trust. Can OpenAI convince customers that more autonomy comes with enough control? GPT-6 Astra has to prove both at once.
The release arrives as OpenAI pushes toward an IPO. The company wants models that businesses trust with harder work. GPT-6 Astra is now the centerpiece of that pitch.
The coming days will put the model in front of a much larger group. That is when the claims meet reality.
Does GPT-6 Astra earn the “AGI era” label, or is OpenAI getting ahead of the evidence? Drop your views below.