OpenAI suspends premium Pro subscription sign-ups amid extreme Astra demand
Infrastructure strain from its new reasoning model forces OpenAI to pause its $200-per-month tier.
Original cover art, generated for this story. THE VISSION does not republish third-party press imagery.
- OpenAI has temporarily suspended new sign-ups for its premium $200-per-month Pro subscription plan.
- Product leader Thibault Sottiaux cited extreme infrastructure strain caused by the launch of its new Astra model.
- Standard lower-tier Plus and Go plans, as well as developer API endpoints, remain unaffected.
OpenAI has officially announced a temporary suspension of new sign-ups for its high-end Pro subscription tier, which costs $200 per month. The company cited a severe, unexpected infrastructure deficit following the public release of its flagship Astra reasoning model. Standard lower-cost consumer plans—including the $20 Plus tier and the mobile-oriented Go plan—as well as developer API endpoints remain open and fully operational, as OpenAI prioritizes existing subscribers and critical commercial integrations.
Thibault Sottiaux, OpenAI’s product lead for consumer applications, confirmed the suspension on Thursday, explaining that Astra’s complex multi-step reasoning and autonomous computer-use capabilities place immense computational strain on the company's clusters. Unlike traditional chat interfaces that consume a few hundred tokens per exchange, Astra actively runs continuous loops of search, self-correction, and tool execution. Sottiaux described Astra as representing a 'generational leap' in computational intensity, requiring OpenAI to aggressively scale its hardware capacity before opening sign-ups to more users.
The subscription pause is a vivid illustration of the 'compute crunch' currently facing frontier AI labs as they transition from static chatbots to active, agentic systems. Because reasoning models consume exponential compute at inference time to 'think' through complex problems, consumer demand can rapidly exhaust available GPU clusters. OpenAI has not provided a definitive timeline for resuming Pro sign-ups, stating only that it is actively deploying new H200 and Blackwell-based server nodes across its data centers to resolve the bottlenecks.
This suspension is a major turning point in the economics of generative AI, demonstrating that inference-time compute is now the primary bottleneck for scaling consumer AI. It proves that even at a premium $200-per-month price point, the hardware costs of running continuous agentic loops can outstrip subscription revenues. Developers and enterprise buyers should expect a prolonged period of pricing volatility and access quotas as labs struggle to balance agentic capabilities with physical chip availability.
How long will the subscription pause remain in effect, and will OpenAI introduce stricter daily token limits for active Astra users to manage capacity?
Still open. When the paper finds out, it will say so here and on the open questions page — including if it got this wrong.