ChatGPT and Codex Will Share One Usage Allowance From December: What OpenAI's Help Center Actually Says

From December, ChatGPT and Codex will share one usage allowance for Plus, Pro and Business plans, measured in tokens. The change is set out in OpenAI's Help Center article "Understanding ChatGPT and Codex usage limits", and it is a small documentation change about how usage is counted — not a new product launch. The ChatGPT Codex shared usage limit means one combined allowance instead of separate allowances for the two services. The exact December day is unstated, and no per-plan allowance size is given.

That distinction matters. The article describes no new model, feature, or price — only how an existing allowance will be counted, "as we bring the agentic capabilities of Work into the main ChatGPT experience". No further detail about that broader move is given, and none is added here.

What OpenAI's Help Center article actually says

The scope sentence names exactly three plans: "Starting in December, Plus, Pro, and Business plans' usage allowance will be shared across ChatGPT and Codex as we bring the agentic capabilities of Work into the main ChatGPT experience." No other plan is named in the text returned. Readers on other plans should not assume the change applies to them.

Usage "will be measured in tokens, the units of information a model processes and generates". How much of an allowance a request uses "will depend on the model you choose and the task you ask it to do", and "more advanced models and complex tasks generally use more of your allowance". No token figure for any model or task type, and no worked example, is given.

One form of use sits outside the allowance: "Instant chat will not draw from your usage." The article does not define Instant chat further, and it lists no other exemption. The safe statement is narrow: Instant chat is exempt. The article also says users "will be able to see your combined remaining usage and reset times in Settings → Usage" — one balance to check, not a separate balance for each service. It does not say when that display appears.

Illustration created for INORYVA — an abstract workspace suggesting chat and coding work sharing one pool, not a screenshot of ChatGPT or Codex and with no OpenAI logo.

At the limit, the article names three options: "you can wait for it to reset, upgrade your plan, or buy credits to continue using eligible features". It does not say which features are eligible, what credits cost, or how long a reset wait would be. It closes with an expectation, not a promise: "We expect most users won't need to wait for their usage to reset more often than they do now." That is OpenAI's forecast for most users, not a guarantee for any individual user.

What the article does not say

First, there is no start day. The article says only "Starting in December", and secondary coverage checked agrees that no specific day has been published. Any source naming an exact December date is adding a date the article does not contain. Second, there are no allowance sizes: no token count, message count, or other size for Plus, Pro, or Business. Secondary coverage agrees the size of each plan's allowance has not been published. No "you will get this much" figure should be quoted as coming from this article.

Third, there is no reset cadence. The article mentions reset times in Settings and waiting for a reset, but states no window — no hours, days, or weekly cycle — for this change. Reset windows from other Help Center articles about other arrangements must not be imported here. Fourth, there is no price information: nothing about any plan's price, a price rise or fall, or the price of credits. A price change must not be inferred from a change in how usage is counted.

Fifth, the date of the Help Center change itself needs care. Secondary coverage reports the article was last updated on 31 October 2026, in US time. That date is secondary-reported only. It did not appear in the primary text fetch used here, which returned the short article text without an update stamp, a date, an allowance size, or a December day. The honest framing is that secondary coverage reports a 31 October update, two days before this article was drafted, while the change takes effect in December.

Illustration created for INORYVA — an abstract single gauge with no numbers, suggesting one shared allowance, not a screenshot and with no invented usage figures.

What one shared allowance means in practice

Under separate allowances, use in one service would not reduce what remained in the other. Under one shared allowance, use in either service draws on the same pool, so heavy use in one leaves less for the other until the allowance resets. That is the direct meaning of "shared" in the scope sentence. It is a statement about counting only: the article does not say the services are merging, that Codex is disappearing, or that ChatGPT gains a named new feature on a named date. The change sits in the same agent-product family as Claude Duet, Anthropic's agent product — but the article claims no parity, and the resemblance ends there.

Planning must stay qualitative until sizes are published or Settings shows a reader's own balance. A short, simple request and a long, complex piece of work will not draw on the pool in the same way, even though the article gives no figures for either. Heavier work will consume more, lighter work less, and the balance is the thing to watch, not a guessed count of requests. The same caution applies to the "most users" expectation: it covers most users in aggregate, before the change takes effect. Readers already close to a limit in either service have the most reason to check the combined display once it appears.

How the change was communicated — and what to check before December

The change reached readers through the Help Center, as far as the searches checked could establish. Searches for an OpenAI blog post about the shared limit returned only Community forum threads, not a blog announcement, and a search of X for an OpenAI post about the shared allowance returned no results. Secondary coverage reached the same finding. That is a searched-absence finding: it describes what these searches found, not proof that no announcement exists anywhere. A separate 28 October X post by Codex lead Thibault Sottiaux, mentioned in secondary coverage, was about bringing Chat and Work together. It was not an announcement of this shared-limit change, and it is not treated as one here.

Illustration created for INORYVA — an abstract settings panel with no readable text or real interface elements, not a screenshot of ChatGPT settings.

Before December, watch Settings → Usage for the combined remaining usage and reset times the article says will be visible there. That display will show a reader's own position — the only individual figure this article can point to without inventing one. No verbatim community reaction is quoted, because no verbatim comment was retrieved. The contrast with fuller announcements underlines how limited this change is: the OpenAI Decisions API launch was a product announcement with its own scope and detail, and ChatGPT picture ads were a product change with visible consequences, while this is a counting rule whose consequences depend on figures not yet published. Do not act on a date or size from secondary summaries until the primary article or the product confirms it.

Key Takeaways

• From December, Plus, Pro and Business share one token allowance across ChatGPT and Codex. The exact day is unstated.
• Use depends on model and task; advanced models and complex tasks generally use more. No per-plan size is published.
• Instant chat will not draw from the allowance. No other exemption is stated.
• Combined remaining usage and reset times will appear in Settings → Usage. No cadence or price change is stated.
• Found in a Help Center article only, as far as searches checked; no blog or X announcement was found. The 31 October date is secondary-reported only.
• OpenAI expects most users will not wait for a reset more often than now. That is a forecast, not a guarantee.

Conclusion

The shared allowance is a narrow, checkable change: one token pool for ChatGPT and Codex, for three named plans, from December, with Instant chat exempt and the combined balance shown in Settings. The day, the sizes, the reset rhythm, and the cost of credits are not in the article that describes it. Until OpenAI publishes them or the product shows them, note the change, watch Settings → Usage, and do not fill the gaps with dates or numbers borrowed from elsewhere. A small Help Center change should be reported at its actual size, not inflated into a launch.

Sources

• OpenAI Help Center, "Understanding ChatGPT and Codex usage limits": https://help.openai.com/en/articles/20001625-understanding-chatgpt-and-codex-usage-limits (fetched 2 Nov 2026, 07:16 PKT).
• 7minai, secondary coverage on the update date and the absence of a specific day and allowance sizes: https://7minai.com/chatgpt-codex-shared-usage-limit-december/ (fetched 2 Nov 2026, 07:16 PKT).
• Automater, secondary coverage on the update stamp and the absence of allowance numbers: https://automater.ai/news/chatgpt-codex-shared-token-allowance-december/ (search-fetched 2 Nov 2026, 07:15 PKT).
• CellCog, secondary coverage on the searched absence of a blog or X announcement: https://cellcog.ai/blog/chatgpt-codex-shared-usage-limit/ (search-fetched 2 Nov 2026, 07:15 PKT).