ChatGPT Pro limits doubled: the numbers OpenAI won't publish

Since 29 October 2026 OpenAI has doubled the allowances for ChatGPT Pro 100, Pro 200 and Pro 500 and raised those of Business Premium. Tibo (@thsottiaux) announced it on X, as BleepingComputer reports. OpenAI's release notes narrow the statement for Pro 200 (per SmartScope; The Decoder reports the same), and none of the sources names a baseline figure.

This article relies on reports by The Decoder (in German), BleepingComputer and SmartScope. We could not fetch the OpenAI pages themselves on 5 November 2026 (HTTP 403 or a dropped connection, a bot wall). What the release notes and the Pro-tiers help article say is therefore given only as SmartScope and The Decoder report it; we have not read the help article itself.

A usage allowance is the amount of work a subscription covers per time window, here for Work and Codex tasks in ChatGPT Pro. According to kingy.ai, OpenAI describes the tiers as multiples of Plus, not as a number of requests or tokens. According to OpenAI (quoted by The Decoder), consumption depends on model, task and settings. Two allowances are therefore comparable only if the provider names the baseline.

Tier	What the sources report	As of	Source
Pro 100, Pro 500	Work and Codex allowance doubled	29 Oct 2026	The Decoder, SmartScope
Pro 200	lower allowance doubled, planned reduction (email of 29 Sep) cancelled	29 Oct 2026	The Decoder, SmartScope
Plus	unchanged	29 Oct 2026	The Decoder
Business Premium	doubling per Tibo's post; per SmartScope the release notes only say "increasing", without a factor	29 Oct 2026	BleepingComputer, SmartScope

Unit of measure: none, because the table contains no usage figures. There is no measurement of our own (no measuring host, nothing billed, no requests counted), and BleepingComputer reports that OpenAI publishes no exact numeric limits per tier. The table only shows who says what, as of our research on 5 November 2026.

What differs on Pro 200

According to SmartScope, the X posts of 30 October (Japan time) spoke of "all usage limits" being doubled. The release notes (entry of 29 October, "More Work and Codex usage on Pro and Business Premium Plans") put it more narrowly, as the "Work and Codex allowance". For Pro 200, SmartScope reports that the lower of the two allowances was doubled, which puts it at the level of the higher one, and that anyone who already had the higher one keeps it. The Decoder reports that on Pro 200 the lower allowance was doubled. The reduction OpenAI announced by email on 29 September is cancelled.

Only kingy.ai names a transition period for existing customers: the Help Center previously listed 29 October 2026 as the transition date, and the page now marks that rule as superseded.

Where the numbers are missing

In late September Tibo named multiples in a follow-up post: Plus 1x, Pro 100 5x, Pro 200 10x, existing customers temporarily 20x. kingy.ai relays this and points out that these are not fixed message counts and that OpenAI has published no numeric weekly meter. The doubling does not make the arithmetic clearer: without a baseline, "twice as much" only says that a value got larger.

The reports give prices in US dollars: 100, 200 and 500 for the three Pro tiers, unchanged according to BleepingComputer. We have not verified euro prices for the EU because we could not fetch OpenAI's pricing pages. They are not listed here.

What this means for purchasing

A company subscription with a fixed price and an unknown allowance has exactly one known number: the price. What an AI model really costs works out cost per task from token price, consumption and wait time, with measured values. OpenAI GPT-5.6: speed as a product covers how OpenAI recently positioned itself on price and speed.

We run no subscription line to OpenAI. Our LiteLLM gateway speaks OpenAI formats, but only against our own models on our own hardware. The doubling changes nothing there, so this is an outside view of the subscription. If you want to evaluate your own pipeline, the topic collection AI in production is a starting point, and what we offer is at Corporate LLM.

Limits of this announcement

Why OpenAI is doing this now is interpretation. The Decoder points to the timing next to Anthropic's upcoming IPO and sees a contest for business customers; that is not a finding. In a later article of 1 November, The Decoder writes that from December, Plus, Pro and Business plans will share one token-based allowance for ChatGPT and Codex (article, in German). The doubling thus concerns an allowance whose framework is due to change from December.

Whether the doubling is permanent is still open; the reports do not say.

Because OpenAI names no baseline, the doubling cannot be confirmed independently. Customers only know that their allowance is supposed to be larger, not by how much.

Correction of 5 November 2026

The first version of 30 October contained statements we could not substantiate, and we removed them: a protection window of 22 to 29 September, euro prices for Pro and Business, a supposed Business Premium description ("5x more usage") on the pricing page, an announced credit, the claim that "Pro 200 gets less", and the statement that the doubling is absent from the release notes (per SmartScope there is an entry about more Work and Codex usage).