OpenAI Canceled the Pro 200 Usage Cut: What Changed for Subscribers

Current purchase correction: The planned Pro 200 reduction is canceled. The September reporting below records the earlier policy; its old eligibility rules are superseded. Current official terms.

Updated September 30, 2026: This article now reflects OpenAI’s revised Pro terms, its September 29 release notes, and Sottiaux’s follow-up. The original publication date and URL are preserved.

Historical source check, October 9, 2026: At that check, OpenAI’s Pro guide described a lower Pro 200 allowance and an October 29 transition for eligible subscribers. That policy is superseded. The 1x, 5x, 10x and 20x descriptions and API-price comparison below remain September reporting, not current published quotas.

What OpenAI has confirmed

Thibault Sottiaux, who leads Codex and ChatGPT Work at OpenAI, previewed the Pro 200 changes in a post on X ahead of OpenAI’s broader announcements. OpenAI has since published the subscription terms in its Help Center.

OpenAI has reopened Pro 200 after pausing new sign-ups and upgrades on September 10. Sottiaux’s September statements described half the old plan’s usage when valued at API list prices. In a follow-up, he gave the then-revised relative allowances: Plus 1x, Pro 100 5x and Pro 200 10x, with eligible existing Pro 200 subscribers temporarily retaining the prior 20x allowance. The Help Center previously listed October 29, 2026 as the transition date.

• No Pro five-hour cap. OpenAI’s Work and Codex pricing guide confirms Pro plans currently have no five-hour limit, although weekly limits may apply.
• Historical transition rule. The September policy protected some existing subscribers temporarily. That distinction is superseded.
• Additional credits. Sottiaux says existing users will receive credits, but has not specified the amount or delivery terms.
• Current dot metering. OpenAI’s current dots guide says conversations with your dot do not count toward ChatGPT usage limits. Tasks it starts or manages in Work or Codex count toward those products’ limits as usual. Deeper work has a plan allowance, with extended limits for the first month after launch.

Sottiaux expects subscribers to complete more work than a month ago as models improve and prices fall. That is OpenAI’s forecast, not a measured result for every subscriber. Model choice and task quality determine whether the revised allowance goes further.

The API-dollar math and its limits

Sottiaux compares the included allowance with the list-price cost of similar work on the API. OpenAI has not disclosed the dollar value of the old or revised allowance. The table is therefore illustrative. A subscription is not a transferable API-credit balance; API-key usage is billed separately.

Illustrative token prices are from OpenAI’s API pricing and model changelog. The table assumes comparable token mixes and proportional metering; it is not a published subscription quota.

The Sol row explains OpenAI’s argument. A lower price can counter a smaller nominal budget for comparable token volume. That does not prove that every user finishes the same amount of work. Long context, tool calls, caching and model capability matter.

Astra did not receive the same announced price reduction. Under the table’s simple assumptions, a smaller dollar-equivalent allowance would buy fewer Astra tokens. The table cannot establish an exact personal quota or task-count cut: OpenAI has not published its numeric weekly meter, and real tasks differ.

Sottiaux’s September follow-up described Pro 200 as 10x Plus, Pro 100 as 5x, and Plus as the 1x baseline while OpenAI raised the Plus floor. He described temporary retention of the prior 20x allowance for eligible existing customers. OpenAI’s Help Center previously described an October 29 transition without publishing those ratios. The historical multipliers are not fixed message counts.

How OpenAI got here

The timeline explains the timing:

• August 25: OpenAI brought the five-hour limit back for Plus in Codex and ChatGPT Work. Both Pro tiers stayed exempt “for the upcoming months.”
• September 3: GPT-6 Astra launched. It shares the Work and Codex allowance rather than having a cap of its own. OpenAI handed out banked resets while access rolled out.
• September 9–10: Sottiaux warned that “demand for Astra is really unprecedented.” The next day OpenAI paused new $200 sign-ups, calling it the smallest step that kept access broad. TechCrunch reported that $200 accounts put the most strain on OpenAI’s systems.
• September 22: GPT-6 Sol and Luna launched at half the price of the 5.6 versions. OpenAI credited improvements to inference and caching.
• September 25: Codex went down. The same day, a long-time Pro 20x subscriber posted on OpenAI’s community forum about hitting a “Limit reached” wall with no explanation of which limit.

The timing suggests OpenAI is balancing demand for expensive models against its subscription capacity. That is an inference from the pause, launch sequence and new limits, not a published accounting of OpenAI’s costs. The clearest public change is the one buyers can verify: $200 remains the price, while the new tier carries a lower included allowance.

Why the API comparison matters

Sottiaux says OpenAI wants API list prices to keep falling as models improve. He argues that a subscription formula based on list prices should pass those reductions through to subscribers, without rewarding OpenAI for keeping public API prices artificially high. The effect depends on which models receive price cuts and which ones a subscriber actually uses.

OpenAI controls both the list prices and the subscription meter. Sol and Luna became cheaper; Astra did not receive the same announced reduction. A buyer cannot derive a personal weekly task count from the 10x label or Sottiaux’s half-dollar comparison alone.

Publishing the weekly allowance in a consistent unit would make the change easier to audit. The Pro terms reviewed for the September reporting described a lower allowance without stating its old or new API-dollar value. Account dashboards can show remaining usage and reset times, but they do not by themselves establish a universal cost per task.

OpenAI’s September 29 release notes add context to the price change. That announcement said dots would not draw on eligible Pro, Business and Enterprise allowances for the following month, after which OpenAI planned to share plan-specific terms. Dots roll out gradually to eligible adults and are unavailable to Pro users in the European Economic Area, Switzerland and the UK at launch. Other new features, including Codex Cloud, Pages, Space and GPT-6.1 Sol, have their own availability and usage rules; OpenAI has not said all of them are exempt from the shared allowance.

How subscribers reacted

The replies to Sottiaux’s first post focused on Astra and on whether current $200 subscribers would immediately lose their allowance. His follow-up addressed the second question in part: eligible existing subscribers keep 20x temporarily and will receive additional credits. The contemporaneous Help Center notice specified October 29 for the old allowance but gave no credit amount. Users whose workloads depend on Astra still need to compare their own before-and-after meter.

What remains unclear

• The numerical allowance. OpenAI has not published an old-and-new weekly API-dollar-equivalent amount or guaranteed task count for Pro 200.
• The additional credits. Sottiaux has not given an amount, delivery date or expiration terms.
• Feature-specific meters. Dot conversations and tasks have different usage rules, as explained above; other features have separate terms. Chat’s GPT-6 Pro model has limits separate from the shared Work and Codex allowance.

What to do if you pay $200

• Check the current terms before changing plans. Compare recent interruptions and the usage display in your account. Use the historical September policy to understand the announcement, not to predict today’s capacity.
• Measure your model mix. The usage dashboard and /status in the CLI show allowance and reset information. Compare repeatable tasks with the same model, speed and scope before drawing conclusions from a rounded meter.
• Keep billing routes separate. API-key use, included subscription usage and purchased ChatGPT credits have different rules. Faster processing can consume included limits more rapidly.
• Compare the current tiers. OpenAI lists Pro 100 at $100, Pro 200 at $200 and Pro 500 at $500 per month. Pro 200 includes more usage than Pro 100; Pro 500 has the highest included usage and includes Astra Ultrafast. At launch, buying credits on Pro 200 does not unlock Ultrafast. Our Pro 500 guide covers that distinct plan.

The September plan reduction described here is superseded. Whether a particular workload shrinks, holds steady or improves depends on the models used and the results they produce. The September reporting left open the numerical weekly meter and the promised credit terms; it does not establish whether credits have since reached a particular account.