OpenAI has temporarily suspended new subscriptions to ChatGPT Pro, its $200-per-month plan, after demand for Astra increased pressure on the company’s infrastructure. The decision applies only to new sign-ups for the Pro tier: the Go and Plus plans remain available for purchase, and access via API continues to be available.
The announcement came from Thibault “Tibo” Sottiaux, head of core products at OpenAI, including ChatGPT and Codex. The rationale provided is straightforward: Pro is the plan that places the heaviest load on systems, and freezing sign-ups was deemed the most targeted move to maintain access for as many users as possible. OpenAI has not provided a date for reopening, nor has it disclosed how many new users were subscribing to the plan each day.
The move underscores how quickly the arrival of a high-demand model can turn into an operational challenge, even for a company operating one of the most widely used consumer AI services. In the generative AI market, product availability does not depend merely on its rollout: sufficient compute capacity is required to support conversations, code generation, reasoning, and other tasks run concurrently by users. When demand surges around subscriptions offering higher access and higher limits, that strain becomes even more visible.
Astra and a launch that accelerated demand
Astra was launched on September 3, and OpenAI is rolling it out across multiple service tiers, including Pro, Plus, Enterprise, and Business, as well as its APIs. The company is framing the model as a major step forward, particularly in reasoning, coding, and computer-use capabilities—three areas where OpenAI faces intense competition and where workloads tend to consume far more resources than brief, occasional queries.
As early as Wednesday, Sottiaux had warned that request volumes for Astra were extraordinary and that OpenAI was adjusting multiple levers to manage the load. At the time, he also floated the possibility of a pause on new Pro sign-ups should demand remain at those levels. The suspension now in place is therefore the execution of a preannounced plan, rather than an abrupt shift in terms of service.
The decision to take action on Pro reflects the structure of the offering. Users on that plan, by definition, have usage terms that place the heaviest demand on the systems compared to lower-cost consumer tiers. Restricting new sign-ups in the most resource-intensive segment allows OpenAI to avoid, at least by its own intentions, a broader shutdown or an across-the-board reduction in access. The company clarified that the priority remains service quality for existing users.
However, no public data is available to assess the exact scale of the issue. OpenAI has not provided figures on overall capacity, the number of Pro subscriptions affected, wait times for prospective subscribers, or any usage thresholds that might shift during this phase. It is therefore impossible to determine whether this measure is set to last just a few days or represents a longer freeze tied to the availability of new compute resources.
The challenge is allocating resources without halting the rest of the service
The decision does not amount to a suspension of Astra nor a shutdown of ChatGPT. Rather, it signals selective capacity management: OpenAI is keeping lower-priced tiers and the API active, while stemming the potential influx of new customers on the tier that concentrates the heaviest resource consumption. For users already subscribed to Pro, the messaging aims to reassure them regarding service continuity, even though potential changes to usage limits have not been detailed.
For those building software products on top of these models, the API remains a key takeaway from the news. OpenAI stated that it remains available, but the episode serves as a reminder that model providers' capacity is a tangible variable for companies and developers: access to a model can hinge on commercial plans, operational priorities, and resource allocation decisions, not just price and performance quality.
The recent context makes the freeze even more significant. Just last month, OpenAI increased usage limits for users of Codex, its dedicated coding tool. That move suggested a phase of greater availability for one of the most computationally demanding categories. The pressure caused by Astra therefore appears recent, at least in the form that has emerged publicly, and tied to the speed at which the new model was adopted following its launch.
The new features related to code and computer use are particularly significant on the cost front as well. An assistant that performs intricate tasks, reasons across multiple steps, or interacts with digital environments can generate longer sessions and more complex requests compared to traditional text conversations. OpenAI has not quantified the specific impact of Astra, but the fact that it chose to curb signups for Pro indicates that the spike is not considered manageable with the current setup alone.
What changes for users and the market
In the short term, anyone intending to upgrade to Pro will have to wait for registrations to reopen or consider one of the other available plans. Go and Plus subscriptions remain open for activation, but OpenAI has not clarified whether they offer equivalent terms for Astra compared to the Pro tier. Organizations operating with Business and Enterprise accounts are also not included in the announced pause, as the suspension exclusively affects new Pro subscriptions.
For OpenAI, the situation highlights a delicate balance. The company must make its models more capable and deliver them quickly to users, yet each product leap can drive costs and demand beyond projections. Keeping multiple access channels open—consumer, professional, enterprise, and API—demands a clear hierarchy of priorities when resources run tight. In this case, the company chose to temporarily sacrifice customer acquisition on its premium plan, aiming to preserve continuity for existing users and the other tiers of its offering.
The situation is also a reminder of the competitive rivalry between AI labs. Astra's claimed capabilities in reasoning, coding, and computer use are precisely what can drive developers, professionals, and enterprises toward a platform. But a technical breakthrough only turns into a full commercial advantage if it can be delivered reliably. Limited access, even when driven by strong demand, can slow adoption among users willing to pay more for priority and higher limits.
The next indicator to watch will be the duration of the pause. OpenAI said it is adding capacity, without specifying a timeline or scale. Reopening Pro subscriptions will be the clearest sign that the company believes it has absorbed the surge generated by Astra. Until then, the pause remains a temporary measure announced with a specific goal: preventing demand for the new model from compromising service for the broader user base.
