OpenAI has turned off new sign-ups for its $200 a month ChatGPT Pro plan. Existing Pro subscribers keep their access. Plus, Go, Business, Enterprise, and the API all still let new customers in.
The company says the Pro plan puts the most strain on its computers, so cutting it off at the door was the smallest change that still lets it serve as many people as possible. The trigger is GPT-6 Astra, OpenAI's newest flagship model, released on September 3. OpenAI's product lead called demand for it unmatched, bigger than anything the company has handled before, including its past periods of steep growth.
There is real substance behind the hype here, which is worth separating from the marketing. Astra is the first OpenAI model to cross into what the company calls a "Critical" risk level for cybersecurity, meaning it is capable of finding serious security flaws that nobody knew existed. That capability is powerful for defense, but it also means OpenAI has to run extra safety checks on top of an already expensive model, which adds more load to its servers.
Cost is the other half of the story. Astra runs at roughly two and a half times the price per question compared to OpenAI's previous top model. A more expensive model to run, combined with a surge of new users wanting it, is exactly the kind of squeeze that forces a company to ration access rather than eat the cost of serving everyone instantly.
None of this is unmatched for OpenAI as a business, even if the demand is. It paused ChatGPT Plus sign-ups in November 2023 after a similar surge and did not reopen them for more than a month. The pattern points to a company whose product demand keeps outrunning its own data center buildout, launch after launch.
For a business reader, the practical takeaway is simple. If you already pay for Pro, nothing changes for you today. If you were about to buy it for your team, you cannot right now, but Plus and the API still work and often cover the same underlying model at a lower cost. Enterprise and Business customers are unaffected, so larger organizations negotiating direct contracts are not stuck waiting behind individual subscribers.
The bigger signal is about where AI pricing is headed. The most capable models are not getting cheaper, they are getting more expensive to run, and the companies selling them are still building out computing capacity to catch up with what they promised. Budget for AI tools with that in mind: the frontier tier will likely keep costing more, even as older, smaller models keep getting cheaper and more available.