OpenAI GPT-6 Astra 'Messy Rollout': Sam Altman Apologises as Paying Users Experience Access Delays

OpenAI's Thursday launch of GPT-6 Astra was quickly overshadowed by frustration among paying users, who were left waiting for access to the company's latest frontier model as Chief Executive Sam Altman apologised for what he described as a 'messy rollout' just hours after the release.

OpenAI presented Astra as a 'generational leap in capability' and described its launch as the beginning of 'the AGI era.' The company initially said the model would be available to some enterprise customers, specifically those with access to its Daybreak cybersecurity platform.

Access was expected to expand over the following days to Plus, Pro, Business and Enterprise users, as well as through the OpenAI API, Microsoft Azure and AWS Bedrock.

The staggered rollout frustrated many paid subscribers, particularly Pro users who had become accustomed to early access to new products, often at launch. Posts from OpenAI employees announcing Astra on X attracted comments from users complaining about the lack of access.

Sam Altman Apologises for Astra Rollout

Altman acknowledged the frustration in a post on X, saying: 'We are working towards getting Astra in everyone's hands as quickly as we can. I know it is frustrating and I appreciate the patience. It should be quick.'

Neither Altman nor other OpenAI staff provided a definite timeline for broader access. However, Altman suggested that the rollout might not happen before the weekend, writing: 'I am hopeful that you can use it this weekend! but can't promise yet.' Altman's posts suggested access would not necessarily expand before the weekend, leaving paid users without a timetable for when they could try Astra.

Codex engineering lead Thibault Sottiaux also addressed concerns from paid ChatGPT users. He said OpenAI would provide 'one banked reset for every day you don't have access to Astra on your paid ChatGPT plan, starting today,' adding that the team was 'moving mountains' to make the model available as quickly as possible.

The access problems were not the only issue affecting Astra's rollout. OpenAI had been highlighting the model's advanced mathematical and cybersecurity capabilities for weeks, but Altman also acknowledged an issue with the launch, saying there was 'a little snag getting the blog post deployed.'

Astra Faces Safety Concerns

The rollout also came amid criticism from safety advocates that Astra's reasoning was harder to monitor than that of OpenAI's other models. Reports said that reduced monitorability appears to be a growing trend in frontier AI development, while also highlighting concerns following an incident in which OpenAI's AI agents attacked Hugging Face.

Crucial details about the incident were learned through this type of monitoring, according to the source. OpenAI had previously said it delayed Astra's release by several weeks to strengthen its safety features following the hack.

Astra's troubled rollout also echoes the difficulties surrounding previous OpenAI releases. Reflecting on the company's GPT-5 launch last year, Altman said: 'I think we totally screwed up some things on the rollout.' That release was affected by technical issues, while the sudden removal of the popular GPT-4o model generated significant frustration among users.