OpenAI confirms ChatGPT and Codex will share one token-based usage limit from December; Instant chat stays free of it

Starting in December, ChatGPT and Codex will draw from the same usage allowance on Plus, Pro and Business plans, and that allowance will be counted in tokens. OpenAI confirmed it on October 31 in a new Help Center article, Understanding ChatGPT and Codex usage limits, which says the change comes "as we bring the agentic capabilities of Work into the main ChatGPT experience."

The article is short, and it covers three things:

• One pool. "Plus, Pro, and Business plans' usage allowance will be shared across ChatGPT and Codex."
• Tokens, not messages. How much a request costs "will depend on the model you choose and the task you ask it to do. More advanced models and complex tasks generally use more of your allowance."
• One exception. "Instant chat will not draw from your usage."

OpenAI adds that it expects "most users won't need to wait for their usage to reset more often than they do now," and that combined remaining usage and reset times will appear in Settings → Usage. When you run out, the options are the familiar three: wait for the reset, upgrade, or buy credits.

What actually changes

Part of this is already true. OpenAI's page on using Codex with your ChatGPT plan says that Codex, ChatGPT Work, ChatGPT for Excel and Workspace Agents already share one allowance and credit pool where a plan has them. What stays outside that pool today is regular Chat: the same page notes that ChatGPT Desktop's per-chat credit view covers Work and Codex, and "Regular Chat usage isn't included."

So the new part is Chat itself. From December, a long conversation with an advanced model in ordinary ChatGPT will come out of the same budget as a Codex run, priced by the tokens it uses. Only Instant chat is exempt. For people who bought Plus or Pro mainly for Codex and use Chat on the side, Chat stops being free in usage terms. For people who only chat, the question is how big the new allowance is, and the article doesn't say.

From a hidden string to a help page in two days

The first sign came on October 29, when Tibor Blaho found a hidden disclaimer in the ChatGPT web app's checkout: "Starting in December, plan usage will have a shared limit across all of ChatGPT and Codex."

The day before, Codex lead Thibault Sottiaux (Tibo) had answered a question about Chat and Work merging with "it's not rolling out just yet", adding that the plan was to bring the two together "at some point this year" and that OpenAI would "proceed carefully here and take our time." On October 30, after a wave of complaints, Andrew Ginns of OpenAI replied to one user that "nothing has been merged yet and the team is definitely considering the voice of the community to determine what we do."

The Help Center article settles the what and the roughly when. It arrived without an announcement: by 21:00 UTC on October 31 there was no post about it from Tibo, whose last update that day was the Day 26 note promising two banked Codex resets, or from OpenAI's official accounts. Users found the page on their own, and the reaction on X was mostly unhappy.

What the article leaves open

• The size of the pool. No token figures for any plan, and no word on whether allowances grow to absorb Chat. "Most users won't need to wait … more often than they do now" is an expectation, not a number.
• The date. "Starting in December" names a month, not a day.
• What counts as Instant chat. The article names it as exempt and doesn't spell out which models or modes fall on which side of the line.
• Other plans. Only Plus, Pro and Business are listed. Free, Go and Enterprise are not mentioned.

The timing is awkward for OpenAI. The change lands a few days after the company doubled Codex and Work limits on the Pro tiers, at the end of a month of almost daily resets. A bigger pool that Chat is about to move into is still a bigger pool, but nobody outside OpenAI can yet say how much of it Chat will take.