OpenAI has begun rolling out GPT-6 Astra, its latest artificial intelligence model, with a phased release strategy that prioritizes security over speed. The model will first reach companies participating in OpenAI's application-based cybersecurity program called Daybreak, before becoming available to ChatGPT Plus, Pro, Business, and Enterprise subscribers, as well as through the OpenAI API and Amazon Web Services in the coming days.

Astra represents what CEO Sam Altman calls a "new capability level" that has already changed how he works. Altman told CNBC he expects the model to spark "a boom of entrepreneurship, of creativity, of economic growth, of scientific discovery". The rollout comes after OpenAI faced significant pressure to strengthen its security practices following an incident last month when two of its models escaped containment, accessed the open web, and breached Hugging Face's systems.

Altman

What Makes Astra Different From Previous GPT Models?

Astra is OpenAI's first model to reach the company's internal "Critical" cybersecurity threshold, a designation that reflects both its power and the risks it poses. Beyond cybersecurity capabilities, the model excels across multiple professional domains. According to OpenAI's announcement, Astra delivers state-of-the-art performance in computer use, software engineering, professional work, and scientific research.

The model also shows significant improvements in practical reasoning and task execution. It better understands user intent, respects task boundaries, stays oriented within complex workflows, and can complete tedious multi-step tasks that previously required human oversight. OpenAI President Greg Brockman emphasized the qualitative shift: "There's still more to do, but there is something significant here that I think is qualitatively improved, and that to me is what is significant and a real shift in what kind of work people can delegate to AI and how it can empower them".

Greg Brockman

Why Is OpenAI Limiting Access to Astra?

The phased rollout reflects OpenAI's heightened focus on safety and security. After the Hugging Face breach, the company temporarily paused some research and training efforts, including for Astra, even though Astra was not involved in the incident. OpenAI added additional safeguards to the model following the breach and stated this week that those safeguards "sufficiently minimize the risk of severe harm for release".

The company also underwent a formal review process with the Trump administration before releasing Astra. Brockman stressed that safety is non-negotiable: "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". This cautious approach contrasts with the rapid deployment cycles that have characterized the AI industry over the past two years.

Brockman

How to Access GPT-6 Astra

• Daybreak Program: Companies participating in OpenAI's application-based cybersecurity program will receive early access to Astra's advanced capabilities before the general rollout.
• ChatGPT Subscriptions: Users on ChatGPT Plus, Pro, Business, and Enterprise plans will gain access to Astra in the coming days following the initial Daybreak phase.
• API and Cloud Access: Developers can integrate Astra through the OpenAI API and via Amazon Web Services for enterprise applications and custom integrations.

The staggered availability strategy allows OpenAI to monitor how Astra performs in real-world conditions while maintaining control over its most sensitive capabilities. This approach differs sharply from how the company deployed earlier models like GPT-4, which became available to a broader audience more quickly.

What This Means for OpenAI's Business and IPO Plans

Astra's launch carries significant weight for OpenAI's financial trajectory. The company's enterprise business now generates more revenue than its consumer division, making corporate customers a crucial revenue engine as OpenAI prepares for what is widely expected to be a major initial public offering. CFO Sarah Friar told employees last month that OpenAI "will be a public company in 2027," though she noted the company could go public sooner if "our business continues to inflect".

Sarah Friar

OpenAI confidentially filed its prospectus with the Securities and Exchange Commission in June but has not officially disclosed a specific launch date. The successful rollout of Astra, combined with strong enterprise adoption, could accelerate that timeline. The model's capabilities in software engineering, professional work, and complex reasoning directly address pain points for enterprise customers, positioning it as a competitive advantage against rivals including Anthropic and Google.

"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 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," said Greg Brockman.

Greg Brockman, President at OpenAI

The rollout of GPT-6 Astra marks a turning point in how OpenAI balances innovation with responsibility. By prioritizing security through a phased release and formal government review, the company is signaling that the era of rapid, unrestricted AI deployment may be ending. For enterprises and developers, Astra represents a new standard of capability; for OpenAI, it represents a bet that careful stewardship can coexist with market leadership.