ChatGPT and Codex Will Share One Usage Allowance From December: What It Means for Plus Users

By Rana — Editor, Elite Era Trends

If you pay for ChatGPT and you also use Codex, your usage is about to be counted differently. OpenAI says that starting in December, the usage allowance on Plus, Pro and Business plans will be shared across ChatGPT and Codex, as it brings the agentic capabilities of Work into the main ChatGPT experience. In plain terms: instead of one meter for everyday chat and a separate meter for coding-agent work, you will have one shared pool — and a heavy afternoon of Codex work will leave less room for chat, and vice versa.

That sounds simple, but the details matter, and OpenAI has not published all of them yet. There is no exact December start date in the Help Center article, and there are no published token amounts per plan. This guide explains what OpenAI has actually confirmed, what it has not said, and how Plus users in particular can plan for December without guessing at numbers OpenAI has not given.

Contents

• What is changing, in one paragraph
• How usage works today — and what changes in December
• Tokens, not messages: why that matters
• Who is affected, and who is not named
• What Instant chat means for your budget
• Three worked planning scenarios (illustrative, not OpenAI figures)
• How to check your usage before December
• What happens when you hit the limit
• What OpenAI has not said yet
• Frequently asked questions
• Sources

What is changing, in one paragraph

Today, ChatGPT Work and Codex already draw on one allowance, while ordinary ChatGPT chats run on their own separate limits. From December, ordinary chat joins that shared pool on Plus, Pro and Business plans. Usage will be measured in tokens — the units of information a model processes and generates — rather than as separate message counts for chat and agent work. You will see your combined remaining usage and reset times in Settings, then Usage. The one exception OpenAI names is Instant chat, which will not draw from your usage.

The change was published in OpenAI’s Help Center article “Understanding ChatGPT and Codex usage limits,” updated October 31, 2026, and confirmed to subscribers by email in early November, as reported by BleepingComputer. OpenAI has not published a separate blog post about it, which is why the Help Center wording is the primary source.

How usage works today — and what changes in December

To understand December, it helps to picture the current setup. If you use ChatGPT Work — the agentic mode that can plan and complete multi-step work across your apps and files — and Codex, the coding agent inside the ChatGPT desktop app, those two already share an allowance. That is why a long Work session today can already leave less room for Codex. What has stayed separate until now is ordinary chat: the everyday conversations where you ask questions, draft emails, summarise documents or brainstorm.

December removes that separation for paid plans. Every non-Instant request in ChatGPT and every Codex run will draw from the same allowance. For someone who only chats, or only codes, the practical effect may be small — one pool instead of two, with the same total work behind it. For someone who does both in the same week, the effect is real: the two kinds of work now compete for the same budget. A coding-heavy week could mean hitting a limit during ordinary chat, and a chat-heavy week could mean less headroom for a Codex task. That competition is the whole story of this change, and it is why planning now beats reacting in December.

What is not changing: Codex is not disappearing. Reporting on OpenAI’s email says it stays a separate experience inside the ChatGPT desktop app, even as the Work tab folds into main chat for a test group first. This change is about counting usage, not removing Codex.

Tokens, not messages: why that matters

The shift from message limits to a token allowance changes how you should think about cost. A token is a unit of text a model reads and writes — roughly, a chunk of a word, not a whole message. OpenAI says how much of your allowance you use “will depend on the model you choose and the task you ask it to do,” and that “more advanced models and complex tasks generally use more of your allowance.”

That has three consequences. First, a short question and a long agent task are no longer comparable like two “messages”: the agent task, reading files, writing code, running tests and revising, can consume far more tokens. Second, model choice matters more: a capable model on a hard task burns the pool faster. Third, you cannot convert old message limits into the new allowance — OpenAI has published no token size, so any exact chat-to-Codex conversion would be guessing. The scenarios below are planning illustrations, not conversions.

For everyday users, the habit to build is checking the Usage screen rather than counting messages — it will be the only reliable measure of where you stand in December.

Who is affected, and who is not named

The Help Center article names three plans: Plus, Pro and Business. Those are the plans moving to the shared allowance. It does not name Free, Enterprise or Edu plans, and reporting has flagged that gap. If you are on one of those plans, do not assume the same change applies to you on the same date — check the Help Center article for your plan, or wait for OpenAI to say more.

The impact is uneven. Plus users who use both chat and Codex are most likely to notice it first, since Plus has the smallest allowance of the three. Pro and Business users start from a larger pool, but the same competition applies. For US small-business owners using AI agents — the audience of our guide to AI agents for small business — this is a budgeting change, not a reason to abandon either tool.

What Instant chat means for your budget

The one carve-out OpenAI names is Instant chat: “Instant chat will not draw from your usage.” OpenAI’s subscriber email describes it as unlimited on paid plans. That makes Instant chat the pressure valve in the new system: quick, everyday questions on the Instant model should not eat into the time you have left for Codex or for heavier agentic work.

There are limits: OpenAI has not said which model counts as Instant once Work folds into main chat, and Instant is a lighter model — not the tool for a complex coding task or multi-step project. Use it for quick questions and save the shared allowance for work needing a capable model or agent. Our GPT-6 Astra pricing guide and GPT-6 Astra for Work pieces explain the model trade-offs.

Three worked planning scenarios (illustrative, not OpenAI figures)

Because OpenAI has not published token amounts, the honest way to plan is with scenarios, not conversions. The three below use a simple, invented unit — call it a “share” of your weekly allowance — to show how the same week can play out differently under one pool. They are planning tools, not OpenAI data, and they deliberately use round numbers so you can replace them with your own Usage-screen figures in December.

Scenario 1: The chat-heavy week. Imagine you spend most of the week in ordinary chat — drafting, researching, summarising — and you run one short Codex task on Friday to fix a small bug. Under separate meters, that Friday task would have had its own budget. Under one pool, the week’s chat has already used most of the allowance, so the Friday task is the one that hits the limit. Lesson: if you know a Codex task is coming, do it earlier in the week, or move light chat to Instant chat first.

Scenario 2: The Codex-heavy week. Imagine you start Monday with a large Codex project — building a feature, refactoring a codebase, running tests — and then try to use ordinary chat for the rest of the week. Under one pool, Monday’s project may leave little headroom for Tuesday’s chat, even for work that has nothing to do with code. Lesson: split large Codex projects across reset windows where you can, and check Settings, then Usage, before starting a second large task in the same window.

Scenario 3: The balanced freelancer week. Chat for client emails in the morning, Codex for delivery in the afternoon — under one pool, neither half has a protected budget, so a heavy afternoon can borrow from the next morning’s chat. Lesson: protect one “must-not-run-out” job weekly, usually the client-facing one, using Instant chat for lighter work and spacing agent tasks around the reset time.

In all three scenarios, the action is the same: check the Usage screen before a big task, not after it fails. That habit costs ten seconds and saves a stalled afternoon.

How to check your usage before December

You do not need to do anything today to keep your plan working, but November is the right month to learn your own pattern. Open Settings, then Usage, in ChatGPT, and look at your combined remaining usage and reset times — that is the screen OpenAI says will show the shared balance in December. If you use the Codex command-line tool, the /status command shows the Codex-side usage today. Check both once a week through November and note two things: which days you come closest to a limit, and whether chat or Codex is usually the one that gets you there. That baseline is the only honest way to judge, in December, whether the shared pool feels bigger, smaller or about the same as the two separate limits it replaces — because OpenAI has not given you numbers to compare against.

Also note your reset times: waiting for a reset is one of OpenAI’s three at-limit options, and scheduling large agent work just after a reset beats starting it just before one. Our explainer on Chrome Autofill and Gmail’s enhanced autofill covers a similar “same tool, new rules” shift.

What happens when you hit the limit

OpenAI lists three options when the shared allowance runs out: wait for the reset, upgrade your plan, or buy credits to continue using eligible features. Credits can be bought from Settings, then Usage, in ChatGPT or from Usage and Billing in the Codex app, according to reporting on the Help Center’s credits guidance, and they last 12 months before expiring. Support does not reset limits on request, so contacting support to ask for an early reset will not help.

Treat credits as a safety net, not a plan. Hit the limit once, and waiting for the reset plus adjusting your Instant-chat mix is usually right. Hit it weekly, and consider upgrading, splitting work across resets, or moving routine tasks to a lighter model. Remember OpenAI’s expectation — “most users won’t need to wait for their usage to reset more often than they do now” — is a forecast without numbers behind it; your December Usage screen is the test.

What OpenAI has not said yet

The gaps, in one place: no exact December start date; no published token size for any plan; no word on whether per-model chat caps stay or go; no mention of Free, Enterprise or Edu plans; no word on how image generation, uploads or regular Chat Voice fit the new allowance; and no word on whether higher-tier subscription credits apply to Codex overflow.

None of that means the change is not happening — the Help Center sentence is unambiguous about the shared allowance itself. It means you should be sceptical of any article that fills those gaps with precise numbers, and you should re-check the Help Center article in December, when OpenAI may update it with the start date and, hopefully, per-plan figures. We will update this guide when it does.

For now: keep using both tools as today, learn your weekly pattern from Settings, then Usage, use Instant chat for quick questions, and do your biggest agent task just after a reset. This is not a price rise or a feature removal — it is one shared budget for two tools. If you use ChatGPT for money work too, our ChatGPT for financial services guide is a reminder of where to keep a human in the loop.

Frequently asked questions

Sources

• OpenAI Help Center — Understanding ChatGPT and Codex usage limits (updated October 31, 2026)
• 7minAI — ChatGPT and Codex will share one usage limit in December
• Skimwire — ChatGPT and Codex to share one usage limit in December
• Limit Digest — ChatGPT and Codex shared limit: OpenAI confirms December

General information only. Confirm current terms in OpenAI’s Help Center and Settings, then Usage, before changing plans.