OpenAI Cuts ChatGPT Pro Usage Limits in Half From October 30

What changes on October 30?

Starting October 30, 2026, the usage included in ChatGPT Pro 200 falls from 20 times to 10 times the ChatGPT Plus allowance ^[2]. The plan still costs $200 a month ^[3]. Existing subscribers keep their current limits through October 29, 2026 ^[3].

The cut applies to ChatGPT Work and Codex. The weekly limit for GPT-6 Pro in Chat also drops from 200 messages to 100 ^[2].

How did the timeline unfold?

• September 10, 2026: OpenAI pauses new sign-ups for Pro 200, citing demand for GPT-6 Astra ^[2].
• At DevDay 2026, OpenAI announces the $500 Pro 500 plan alongside the cut ^{[2] [3]}.
• September 30, 2026: Pro 200 reopens to new subscribers with roughly half the usage ^[2].
• October 29, 2026: the last day of current limits for existing subscribers ^[2].
• October 30, 2026: the reduced limits take effect ^[2].

How much less can you use?

PCWorld describes the change as a 50 percent cut to the plan's usage over Plus. The $200 plan will now only offer 10 times the usage limits of Plus ^[1]. Existing Pro subscribers get a brief grace period and usage credits to soften the change ^[1]. The sources here do not give a credit amount, so the value of those credits is unknown. The grace period is described as brief, and no source gives its exact length ^[1].

OpenAI has also said it will not bring back the five-hour usage window on this plan ^[3]. Users can spend their weekly allowance whenever they like ^[2], which helps people who work in long sessions.

The 10 times figure comes from a rough chart that OpenAI's Thibault Sottiaux shared on X ^[1]. None of the reports here links to an OpenAI help page that states it, so treat the exact numbers as the best current estimate and check OpenAI's official pages before you plan around them.

Why is OpenAI making the cut?

OpenAI paused new sign-ups for Pro 200 on September 10 because of demand for GPT-6 Astra. It reopened the plan on September 30 with roughly half the usage ^[2].

Thibault Sottiaux, OpenAI's Head of Core Products according to PCWorld, pointed to new DevDay features that demand a lot of compute ^[1]. He also said OpenAI passed on other ways to compromise, including a rolling five-hour limit ^[1]. Sottiaux said the new calculation nets out to half the dollar value in API spend compared with the old plan ^[2].

The company says ChatGPT now has 1.2 billion weekly users ^[3]. Sottiaux promised that subscribers would get more work done over time as models become more efficient ^[1].

What does this mean in practice?

The price stays the same while the allowance halves ^[2]. A heavy Codex user will reach the weekly limit after roughly half the work they did before ^[2].

The reports here do not describe a change to the allowance for Plus subscribers ^[2].

Three effects stand out:

• Plan large Codex or ChatGPT Work jobs around a weekly allowance that is half the size.
• Some subscribers are leaving. PCWorld reports that some users are cancelling their subscriptions in protest ^[1].
• The old limits end soon. Current limits last through October 29, so work that depends on the old allowance should be finished or moved before then ^[2].

PCWorld reads the net effect as essentially a wash, because OpenAI says newer models get more done with fewer tokens ^[1]. That view depends on the models becoming more efficient, which the sources do not show for your own workload.

Should you move to the $100 or $500 plan?

OpenAI's Pro lineup now has three tiers: $100 at 5 times Plus, $200 at 10 times Plus for new subscribers, and $500 at 25 times Plus ^[2]. OfficeChai works out that each tier comes to about $20 per Plus-equivalent unit, so the bigger plans no longer give a volume discount ^[2].

Monthly plan	Usage compared with Plus	Cost per Plus-equivalent unit
$100	5 times	about $20
$200 (Pro 200, new subscribers)	10 times	about $20
$500 (Pro 500)	25 times	about $20

The table uses the figures in OfficeChai's report ^[2]. The per-unit cost is its arithmetic, not a price OpenAI publishes.

The $500 Pro 500 plan is the only Pro tier with Ultrafast, which can generate up to 300 tokens per second in Codex and which OpenAI says is up to 8 times faster than standard generation ^[3]. GPT-6 Astra Ultrafast is available now, and GPT-6.1 Sol Ultrafast is coming later ^[3]. Ultrafast is only on Pro 500, so speed in Codex is one reason to pay for that tier. The 25 times allowance is the other. Otherwise, the choice comes down to how much usage you need, not which tier looks cheapest.

Are there other ways to get the work done?

Some activity does not count against the subscription. Sottiaux hinted at "more things to the subscription that won't draw on the usage", and PCWorld reports that this included chats with OpenAI's new Dot assistants ^[1].

For developers who can pay per use, the API is the other route. PCWorld notes that Sottiaux pointed to a recent 50 percent API rate cut for GPT-6 Sol and GPT-6 Luna ^[1]. API rates are pay-as-you-go, not a flat subscription rate ^[1], so the cost follows the work you do rather than a fixed weekly allowance.

Will other AI subscriptions change too?

PCWorld expects more of the same: AI providers keep changing flat-rate plans as agent features use more tokens. It writes, "Expect more of the same in the weeks and months to come" ^[1]. That makes it worth checking every few months which features in your plan draw on the allowance and which do not, because the rules can shift again with the next model or feature launch. Read OpenAI's official pages on October 29, the last day of current limits, so you see the new numbers before they apply.

Before October 30, estimate a typical week of your own use. If 10 times Plus does not cover it, the $500 tier is the only Pro option with a larger allowance.