On October 29, 2026, OpenAI doubled usage limits for Work and Codex on ChatGPT Pro 100, Pro 200, Pro 500, and Business Premium plans: the update took effect immediately and was automatically applied to active subscriptions. Simultaneously, the planned reduction of limits for Pro 200 starting October 30 was canceled, so the increased quota is in effect at the previous subscription price. The Plus plan did not receive any changes. OpenAI warns, however, that a doubled quota does not mean twice as many completed tasks, and Fast mode consumes limits significantly faster than standard mode.

What happened

The change was announced by Tibo Sottiaux (@thsottiaux), head of the Codex and ChatGPT divisions at OpenAI; the announcement was presented as "day 25 of 28 days of improvements." Starting October 29, Work and Codex limits were doubled for Pro 100, Pro 200, Pro 500, and Business Premium plans, and the adjustment was automatically applied to existing subscriptions: nothing needs to be activated, and the new values are already displayed on the usage page. For Pro 200, the adjustment is twofold: in addition to the doubled quota, the planned reduction of limits starting October 30 was canceled, which OpenAI had warned about in a letter dated September 29. Subscribers who had already been switched to the reduced limit returned to the increased quota for the previous $200 per month. The Plus plan was not affected by the adjustment.

Context

The chronology of recent weeks changes the perception of the news. On September 29, OpenAI warned Pro 200 subscribers in a letter that their limits would be reduced starting October 30; this cut, according to widespread assessment, triggered a developer exodus, so the October 29 announcement is interpreted as a direct response to it and as a maneuver in the intensifying battle for the corporate B2B market ahead of Anthropic's IPO — the latter interpretation is a working interpretation, not officially confirmed OpenAI motives. The verifiable part of the story is drier: the letter about the reduction from September 29 turned into the cancellation of the reduction on October 29, meaning that in about a month, the limit policy changed to the opposite. The conclusion from this chronology is simple: subscription quotas are a reactive variable that moves in response to consumption telemetry and market conditions, and it is reasonable to factor this understanding into pricing plans.

Why this matters for the industry

Materially, the changes affect the economics of agentic development, not the technology: neither the models nor their capabilities changed — what changed is how much work fits into a subscription. With unchanged prices, including $200 per month for Pro 200, the doubled included quota makes a unit of limited computational capacity approximately twice as cheap: agentic workflows that previously did not fit into the plan are now executed without extra charges, and the barrier to intensive daily work with Codex and ChatGPT Work is lowered. The second effect is competitive: not every provider can ensure such output at the same price without losing margin, and against the backdrop of Anthropic's upcoming IPO, subscription quotas are becoming a field of price competition and a tool for retaining developers. If the doubling turns out to be a trend rather than a one-time action, one should expect counter-moves from competitors and further revisions of limits in both directions.

Why this matters for users

For paid subscribers of Pro 100, Pro 200, Pro 500, and Business Premium, the doubled Work and Codex limits are already in effect: activation is not required, and the new values are visible on the usage page. Pro 200 subscribers additionally avoided the cut scheduled for October 30: the quota increased at the previous $200 per month, and those who had been switched to the reduced limit returned to the increased one. It is reasonable to spend the limit reserve as follows: perform long tasks in Standard mode, and leave Fast and Ultrafast for latency-sensitive scenarios, because Fast mode consumes the quota 2.5 times faster than standard mode. It is also worth reviewing your own processes that previously had to be squeezed due to limits: some of them now fit entirely into the subscription. These changes do not apply to the Plus plan.

What is currently unknown / limitations

Two clarifications are important for an honest assessment. First — from OpenAI's own announcement: doubling the quota does not guarantee twice as many completed tasks, useful output grows non-linearly, and Fast mode consumes limits 2.5 times faster than Standard, so a "50% actual discount" is correct as a reduction in the price per unit of included quota, but not as a measured increase in useful work. Second — the connection between the step and the developer exodus after the September cut and with Anthropic's IPO remains an interpretation, not confirmed by OpenAI data. This is compounded by the question of sustainability: the warning from September 29, the cut scheduled for October 30, and its cancellation on October 29 show that limits are revised quickly and in both directions, so it is advisable to build the current quota into long-term plans with a buffer. Competitor reactions and the content of the remaining days of the "28 days of improvements" campaign are currently unknown.

Sources

Author

Look at AI, editorial team