Simply hours after OpenAI launched GPT-6 Astra, CEO Sam Altman was already apologizing for what he describes as a “messy rollout” after paying customers anticipating entry to the brand new frontier mannequin had been left ready.
The corporate hailed the mannequin as a “generational leap in functionality” on Thursday and described it as the beginning of “the AGI period,” a fuzzy and poorly outlined time period tech executives nevertheless insist on using to advertise their merchandise. It said Astra would roll out to some enterprise clients — particularly these with entry to its Daybreak cybersecurity platform — that day. This is able to develop to incorporate all Plus, Professional, Enterprise, and Enterprise customers, in addition to via the OpenAI API, Microsoft Azure, and AWS Bedrock, over the following few days.
The staggered launch that appeared to prioritize enterprise clients irritated a lot of OpenAI’s paid subscribers, notably these on the dearer Professional plan accustomed to having access to new merchandise first, often at launch. Social media posts from OpenAI workers announcing the launch on X are crammed with feedback complaining about entry.
“We’re working in direction of getting Astra in everybody’s arms as rapidly as we are able to,” Altman stated in an X post acknowledging the frustration. “I do know it’s irritating and I admire the endurance. It must be fast.”
Neither Altman nor different OpenAI workers on-line offered a timeline for when entry to Astra could be expanded, although Altman’s posts counsel this is not going to occur earlier than the weekend, and probably not even then. “I’m hopeful that you need to use it this weekend! however can’t promise but,” he wrote.
“We are going to give one banked reset for each day you don’t have entry to Astra in your paid ChatGPT plan, beginning at this time,” said Codex engineering lead Thibault Sottiaux in a bid to calm considerations. “Workforce is shifting mountains to offer entry as quick as we are able to.”
There have been additionally points with different facets of Astra’s rollout, which the corporate has been teasing with updates on its advanced mathematical and cybersecurity capabilities for weeks. Altman said there was “slightly snag getting the weblog put up deployed.” OpenAI has additionally confronted intense criticism from safety advocates after it stated Astra’s reasoning was harder to watch than its different fashions. This seems to be a rising development on the frontier of AI growth however thought of particularly regarding in mild of OpenAI’s AI brokers attacking Hugging Face, about which crucial details were learnt via this sort of monitoring. OpenAI stated it delayed Astra’s release by numerous weeks to bolster security options within the wake of the hack.
Different huge OpenAI releases have additionally been marred by missteps and frustration. “I feel we completely screwed up some issues on the rollout,” Altman said final yr when reflecting on the corporate’s final main launch, GPT-5. The launch was plagued with technical points and the choice to instantly take away the favored GPT-4o mannequin sparked outrage amongst customers.
