# OpenAI will pause $200 Pro subscriptions as Astra strains capacity > Source: > Published: 2026-09-10 19:22:57+00:00 # OpenAI will pause $200 Pro subscriptions as Astra strains capacity **Existing subscribers keep access, while OpenAI leaves its other plans and metered API open as it adds compute.** By [Ryan Merket](/author/ryan-merket) ยท Published Primary source: [X](https://x.com/thsottiaux/status/2098113585683808624) ## Why it matters OpenAI is using admission control to protect Astra capacity just one week after launch, exposing the tension between compute-heavy agents and fixed-price subscriptions. OpenAI will pause new subscriptions to its $200-a-month ChatGPT Pro plan after heavy use of GPT-6 Astra put more pressure on its systems than other subscription tiers, according to [Thibault "Tibo" Sottiaux (@thsottiaux)](https://x.com/thsottiaux/status/2098113585683808624), the OpenAI executive who leads Codex. Sottiaux said on September 10th that existing Pro subscribers would retain their accounts and access to Astra. OpenAI's other subscriptions and API remain available. He described suspending the $200 tier as the smallest intervention that would preserve broad access while OpenAI adds capacity. "These put the most strain on our systems," Sottiaux wrote, referring to $200 Pro subscriptions. The decision turns OpenAI's highest-priced individual subscription into an admission-controlled product seven days after Astra's release. Sottiaux had warned a day earlier that OpenAI might stop accepting new Pro subscribers. In [that September 9th post](https://x.com/thsottiaux/status/2097559315150426222), he called demand for Astra unprecedented and said OpenAI would prioritize service for existing customers. The speed of the restriction matters. OpenAI [released GPT-6 Astra on September 3rd](https://openai.com/index/gpt-6-astra/) and began rolling it out across ChatGPT Plus, Pro, Business and Enterprise, along with the API, Microsoft Azure and AWS Bedrock. OpenAI said subscription users could consume Astra through their existing usage allowances, while Pro, Business and Enterprise customers would also receive access to a higher-compute Astra Pro mode. OpenAI designed Astra for long-running, compute-heavy work spanning coding, computer control, research and document creation. In Codex, Astra can execute multistep software-engineering jobs and retain notes across context windows. Those capabilities encourage the sustained sessions that make a fixed monthly subscription harder to provision than ordinary chat traffic. The $200 tier concentrates that load among customers who bought it specifically for higher usage. Suspending new sales protects capacity for people already paying while leaving OpenAI's metered channels open. Developers can still use [GPT-6 Astra through the API](https://developers.openai.com/api/docs/models/gpt-6-astra), where standard pricing is $10 per million input tokens and $50 per million output tokens. Prompts exceeding 272,000 input tokens carry higher rates. That distinction gives OpenAI tighter control over the economics of additional demand. API customers pay according to consumption, while individual subscribers pay a recurring fee for an allowance that can support expensive agent runs. Sottiaux's statement identifies the $200 accounts as the largest source of system strain, despite their higher monthly revenue. OpenAI has not withdrawn Astra from its wider rollout. OpenAI's launch materials said Astra usage would count against existing subscription allowances, with customers able to purchase credits for additional use. The subscription pause instead limits the number of new heavy users entering the system while OpenAI brings more compute online. Sottiaux has become the public operator for that rollout. An [OpenAI Forum biography](https://forum.openai.com/public/events/codex-is-for-everyone-why-codex-matters-beyond-code-fa40puy7wi) identifies him as the leader of Codex. Before OpenAI, he led human-data work for Gemini and built AI and machine-learning workflow infrastructure at Google DeepMind. He previously worked as a Google software engineer and studied computer science, computational mathematics and applied mathematics at Universite catholique de Louvain. His role makes the subscription decision an operating response rather than a routine pricing change. OpenAI is keeping existing contracts intact, preserving lower tiers and directing incremental demand toward products where usage is easier to meter. The pause shows how quickly a flagship model's distribution can become constrained by inference capacity, even when access is sold through OpenAI's most expensive individual plan.