AI · 29 October 2026 · 4 min read
OpenAI said it would double the Pro limits. The next sentence says one tier stays the same
On 29 October, OpenAI’s Thibault Sottiaux posted that Pro 100, Pro 200, Pro 500 and Business Premium usage limits are doubled immediately, and that the planned cut to Pro 200 is cancelled. The same post says Pro 200’s Work and Codex allowance will stay as it is. The post does not give a number of messages.
OpenAI had told Pro subscribers that the $200 plan would include less. From 30 October, the included use of ChatGPT Work and Codex on Pro 200 was due to fall from 20 times the Plus allowance to 10 times, and GPT-6 Pro messages in chat from 200 a week to 100. Existing subscribers were to keep the old ceiling until 29 October, plus a one-time grant of usage credits. The price was not changing. A cut on a fixed bill is a quieter price rise. That was the plan published at the end of September.
On 29 October the company reversed it in public, and the wording does not quite agree with itself. Thibault Sottiaux, who speaks for ChatGPT and Codex, posted that OpenAI is doubling usage limits for Pro 100, Pro 200, Pro 500 and Business Premium, effective that day. The next sentences say the company heard the feedback and will not reduce Pro 200, “so your Work and Codex usage will stay the same.” Doubled, and unchanged, cannot both describe the same allowance. A reader of the post cannot tell which sentence governs the $200 plan. The $100 and $500 plans, and Business Premium, are only in the first sentence.
What the post does not contain is a number. No messages per week, no tokens, no multiple of the Plus plan. The September note had those. The October note replaces them with “doubling” and “stay the same,” and adds that extended limits for Dots, the always-on assistant, continue for now and still do not draw on the main allowance. A limit that is not printed cannot be checked against a bill. Until the help page states the new multiple, the only firm fact is that the 30 October cut, as described in September, is not going ahead.
The argument underneath is the one a heavy user already knows. Work and Codex share an allowance. A coding agent that runs for an afternoon can spend a week’s chat. OpenAI’s answer, this week, was to leave the price cards where they are and move the ceiling, after people objected. That is a product decision, not a model release. It does not say the computers got cheaper. It says the company would rather meter less strictly than lose the subscribers who were about to hit the wall.