OpenAI pauses new sign-ups for its $200 ChatGPT Pro tier, citing demand for GPT-6 Astra it can't yet serve
OpenAI says compute demand from Astra, which reached general availability on September 3, has outrun the capacity it has to support new $200-tier subscribers -- with no date given for when sign-ups reopen.
OpenAI paused new sign-ups and upgrades to its $200-a-month ChatGPT Pro tier ("Pro 20X") effective September 10, according to CIO. Thibault Sottiaux, a member of OpenAI's technical staff, said the move is meant to protect service for people already on the plan: "To make sure our current users have an incredible experience and continued access to Astra, we are going to pause subscriptions to our $200 Pro plan." He added: "Demand for Astra is really unprecedented." Existing $200 subscriptions keep renewing, and the $100 Pro tier, along with Go, Plus, and API access, remains open to new subscribers.
A capacity problem tied to how Astra gets used
GPT-6 Astra, OpenAI's flagship model, reached general availability on September 3 and was notable at launch for becoming OpenAI's first model to cross the "Critical" cybersecurity capability threshold under its own Preparedness Framework -- coverage this site published at the time. What's new here is the demand side: OpenAI has not published usage figures, but the pause itself, combined with Sottiaux's comment that Astra's computer-use feature -- letting the model operate a desktop the way a person would -- draws more compute per session than earlier models, points to a specific bottleneck rather than a general capacity shortfall. Anyone who cancels or downgrades out of the $200 tier during the pause loses the ability to resubscribe to it until OpenAI lifts the restriction.
No timeline, and a pattern worth watching
OpenAI gave no date for when Pro 20X sign-ups will reopen, saying only that it is working to add capacity. That's a notable admission from a company that, a week earlier, was promoting Astra's broad rollout across ChatGPT's paid tiers, the API, and AWS. For a site tracking whether AI spend converts into usable capacity rather than announcements, the gap between an advertised general-availability launch and an ability to actually onboard new high-tier subscribers eight days later is the more informative number than any benchmark score in the original release.