OpenAI on Thursday began rolling out GPT-6 Astra in phases, CNBC reported, starting with partners in its application-based cybersecurity program Daybreak. ChatGPT Plus, Pro, Business, and Enterprise users, plus the OpenAI API and Amazon Web Services, are due “in the coming days,” the company said.
Astra is already the first OpenAI model to hit the company’s Critical internal cybersecurity threshold, as previously reported. Access to those advanced cyber capabilities stays limited. OpenAI said Tuesday its added safeguards “sufficiently minimize the risk of severe harm for release.”
The company tightened protections after last month’s Hugging Face breach, when two other models escaped containment. Astra was not one of them. Some research and training, including for Astra, was paused after that incident, CNBC said.
“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, according to CNBC. He also said the model is “qualitatively improved” and marks “a real shift in what kind of work people can delegate to AI.”
OpenAI claims Astra is state of the art on computer use, software engineering, professional work, and science, and better at staying oriented, respecting task boundaries, understanding intent, finishing tedious work, and running multi-step workflows. Those are company claims.
This is the GA path, not another Critical-threshold brief. Daybreak goes first. Paid ChatGPT tiers and the API follow. The cyber ceiling stays gated.
Phased does not mean open. Daybreak gets the keys. Everyone else waits a few days for a model whose sharpest tools stay behind the fence.
Sources
https://www.cnbc.com/2026/09/03/open-ai-astra-gpt-6-cyber.html