Nov 1, 2026

ChatGPT and Codex Will Share One Usage Limit in December: What Counts, What Doesn't, How to Prepare

Starting in December, if you pay for ChatGPT Plus, Pro or Business, your chats in ChatGPT and your coding tasks in Codex will draw from one shared allowance, measured in tokens. Instant chat will not count against it. You don’t need to do anything today, but if you use both ChatGPT and Codex heavily, start checking Settings → Usage every week now, so that in December you can tell whether long chats are eating into the time you have left for Codex. OpenAI says it expects most users “won’t need to wait for their usage to reset more often than they do now.” It has not published the size of each plan’s allowance or an exact start date. Everything below needs no code.

All of this was checked on November 1, 2026 (Beijing time) against OpenAI’s new Help Center article Understanding ChatGPT and Codex usage limits, last updated on October 31, 2026 (US time). We also used Using Codex with your ChatGPT plan, ChatGPT Work and Codex, Using credits for flexible usage (personal plans) and OpenAI’s Codex pricing page.

Key facts:

Who: Plus, Pro and Business plans. (OpenAI Help Center)

When: “Starting in December.” OpenAI has not published a specific day.

What changes: one allowance shared across ChatGPT and Codex, measured in tokens. How much you use depends on “the model you choose and the task you ask it to do.”

What does not count: “Instant chat will not draw from your usage.”

Where to check: Settings → Usage will show your combined remaining usage and reset times.

If you run out: wait for the reset, upgrade your plan, or buy credits.

Why: OpenAI says it is bringing “the agentic capabilities of Work into the main ChatGPT experience.”

What is changing, in plain words

Today your ChatGPT plan has two separate buckets:

Chat limits. Everyday ChatGPT conversations have their own limits. For example, Pro tiers have a weekly message limit for GPT-6 Pro in Chat (see our Pro usage update).
An agentic allowance. Codex, ChatGPT Work, ChatGPT for Excel and Workspace Agents already share one allowance and one credit pool when your plan includes them (Using Codex with your ChatGPT plan). On the Codex side it has a 5-hour window and a weekly window, and you can see both resets in the usage dashboard.

From December, OpenAI says your plan’s “usage allowance will be shared across ChatGPT and Codex,” and your usage “will be measured in tokens, the units of information a model processes and generates.” Put simply: the conversations you have in ChatGPT and the coding work Codex does for you will both come out of the same allowance. The only exception OpenAI names is Instant chat.

If you also use Claude, it already works this way: Anthropic says chatting on claude.ai or Claude Desktop uses the same allowance as Claude Code (see our Claude Code usage limit guide).

What a “token” means for you

A token is a small piece of text. Your message, any files you attach, the earlier messages in the chat, tool results and the answer all count. OpenAI’s pricing page says prompt length alone “isn’t a reliable estimate.” Model choice, context, reasoning, tool use, retrieval and caching all change how much a task uses (Codex pricing). In practice, three things use the most:

Bigger models. OpenAI says “more advanced models and complex tasks generally use more of your allowance.”
Long chats. Every new turn in a long chat sends the earlier context along with it.
Agent work. A long-running task “can use substantially more than a short request,” because the agent runs tools, reads files and makes many model calls (Codex help).

What will and won’t count

Here is what OpenAI’s pages say today. Items marked “not stated” are things OpenAI hasn’t said yet. Don’t assume either way.

Activity	Counts toward the shared allowance from December?	Source
Instant chat	No, “Instant chat will not draw from your usage”	Usage limits article
Codex (desktop app, CLI, IDE extension, web, Codex Cloud)	Yes	Usage limits article
ChatGPT Work, ChatGPT for Excel, Workspace Agents	Yes, these already share Codex’s allowance	Codex help
Other ChatGPT chat (non-Instant models)	Yes, it is “shared across ChatGPT and Codex”	Usage limits article
Codex auto-review safety checks	No, free when signed in with ChatGPT	Codex help
Image generation, file uploads, regular Chat Voice	Today they have separate limits; how they fit into the new allowance is not stated	Codex help
Codex with your own API key	No, billed at API rates instead	Codex pricing
Free and Go plans	Not mentioned in the announcement	Usage limits article
Enterprise and Edu	Not mentioned; they have their own credit or USD billing	Codex help

Two things are still unknown: how big each plan’s allowance will be and the exact December date. OpenAI says it is “preparing carefully for this transition to help prevent disruption.” Codex lead Thibault Sottiaux also wrote on X on October 28 that the team is bringing Chat and Work together carefully and will share more soon, as reported by ABAB News. We will update this guide when OpenAI publishes the numbers.

Step 1: Find your usage page now

You want to know where your usage stands today, so you can tell in December whether the change hurt.

In ChatGPT (web or desktop app):

Open Settings from your profile menu.
Select Usage.
Note your remaining usage and the reset times shown. OpenAI says this is where “your combined remaining usage and reset times” will appear after the change.

In Codex:

In the Codex app, open Usage & Billing. This is also where you can see your credit balance and buy credits (credits help).
The web dashboard is at chatgpt.com/codex/settings/usage.
If you use Codex in a terminal, type /status in a running session to see what you have left. If you’d rather not type commands, ask Codex: “Show me my remaining usage.”

OpenAI’s own advice is to “check the dashboard every week or two to understand your pace and remaining capacity” (Codex pricing). Write down how far through your weekly allowance you usually are by Friday. That gives you something to compare against in December.

Step 2: Make the allowance last

These habits use only settings OpenAI documents. None of them need code.

Use Instant for everyday questions

Instant chat is the one thing OpenAI says will not draw from the shared allowance. Use it for quick lookups, rewording an email, translating a paragraph or simple explanations. Save the bigger models for work that really needs them, and you keep more of the allowance for Codex.

Pick a lighter model for routine Codex tasks

In Work and Codex, the model picker lets you choose Default or a specific model. Default moves models and reasoning levels along one slider (ChatGPT Work and Codex). OpenAI’s pricing page shows how much the model matters. For Plus, it estimates these local Codex messages per five-hour period (Codex pricing):

Model	Plus: local messages per 5 hours (estimate)
GPT-6 Astra	5–45
GPT-6.1 Sol	15–160
GPT-6 Sol	15–150
GPT-6.1 Terra	35–285
GPT-6 Luna	350–3,000

OpenAI says these “are not fixed message limits,” and cloud tasks may use more. The pattern is still clear: Terra gives you roughly twice as many messages as Sol, and Luna many times more. Our GPT-6.1 Terra guide shows where to switch models in ChatGPT and Codex. Codex keeps a model you choose manually until you change it.

Turn Fast mode off unless you need it

Speed modes use your included allowance faster. Fast uses it at 2.5x the Standard rate. Ultrafast on GPT-6 Astra and GPT-6.1 Sol uses it at 8x, and Ultrafast is only on Pro $500 (Codex pricing). If you’re not waiting on the answer, Standard gets you the same work for less of your allowance.

Start fresh chats instead of endless ones

Long chats carry their whole history with them. When a task is done, start a new chat with a two-line summary. OpenAI recommends the same if Codex keeps showing “Context automatically compacted” without making progress: stop the task and start a new chat with a short summary of your goal and what’s done. A new chat does not reset your limits (Codex help). It just stops the waste.

Be careful with images

On the Codex side, image generation uses included limits “~3-5x faster on average” than similar turns without images (Codex pricing). OpenAI hasn’t said yet how ChatGPT’s own image limits will work after the change.

Give agents tight instructions

A vague request like “clean up this project” can run for a long time. Tell Codex or Work exactly which files to touch, what “done” looks like, and to stop and ask before anything big. Short, specific tasks are cheaper than open-ended ones.

Step 3: What to do when you hit the limit

OpenAI lists three options: “wait for it to reset, upgrade your plan, or buy credits to continue using eligible features.” Here’s how each works today.

If you’re in the middle of a Codex task: Codex can usually finish the turn it’s working on, “subject to fair-use limits.” After that, the limit banner or usage page shows your options (Codex help).

Wait. Check the reset time on Settings → Usage. Don’t contact Support to ask for a reset. OpenAI says plainly that Support “does not reset ChatGPT or Codex usage limits.”

Buy credits (Plus and Pro). You don’t have to change plans. Buy them from Settings → Usage in ChatGPT or Usage & Billing in the Codex app. Before you buy, know that (credits help):

Your plan’s included usage is used first, then credits.
Credits last 12 months and then expire. They don’t roll over.
Credits are generally non-refundable and can’t be transferred.
Credits work across supported features. For example, credits you bought in Codex can pay for ChatGPT Work.
Your credit balance is separate from your ChatGPT wallet. Redeeming a gift card does not add credits.
If a task finishes after your balance runs out, the balance can go negative. Your next purchase pays that off first.
Automatic reload (if offered to you) tops your balance back up whenever it drops below a minimum. You can set a maximum monthly spend. Turning it on when you’re already below the minimum can trigger a purchase straight away. If you’re trying to keep costs down, leave it off.

Use a banked reset if you have one. Some Plus and Pro referral promotions give you a “banked” Codex reset. Open the profile menu and select the usage summary showing something like “1 reset available.” A full reset refreshes your 5-hour and weekly Codex windows and changes your weekly reset date (Codex help).

Upgrade. Pro comes in $100, $200 and $500 tiers (Codex pricing). OpenAI doubled the Work and Codex allowance on Pro 100 and Pro 500 on October 29 and canceled the Pro 200 cut (our Pro update). Wait for OpenAI’s December numbers before you upgrade only because of this change.

Should you change anything before December?

You mostly chat and rarely use Codex: probably nothing to do. Use Instant for everyday chat and keep an eye on Settings → Usage after the change.
You use Codex a lot and chat a lot on bigger models: this is the group most likely to notice. Start checking usage weekly now, move everyday questions to Instant, and use Terra or Luna for routine coding.
You were thinking of buying credits in bulk: remember they expire after 12 months and are generally non-refundable. Buy small amounts until you’ve seen how December works for you.
You run Codex in scripts or CI with an API key: that’s billed at API rates, not your plan, so this change doesn’t affect it (Codex pricing).
You’re on Business: Business is named in the change. Ask your workspace admin whether your workspace has purchased credits. With credits, people can keep working after their included limit runs out (ChatGPT rate card).

Troubleshooting

“My usage dropped faster than expected after one task.” Look at the model, the speed mode and how long the chat had become. Astra and Fast or Ultrafast use far more than Terra or Luna at Standard. For Plus and Business Standard seats, Astra usage is limited, with optional credits after that (ChatGPT Work and Codex).

“I bought credits but can’t see them.” First make sure you’re in the right account or workspace, then check recent usage. A negative balance from an earlier task can soak up a new purchase (Codex help).

“The numbers look wrong.” Contact OpenAI Support with the limit or balance you’re questioning, the reset time shown, a screenshot, the Codex client and model, and when it happened, including your time zone. Remove sensitive information from the screenshot first (Codex help).

“I see an image or upload limit banner in ChatGPT. Will that stop Codex?” Today, no. Banners like “50 images in the last day,” voice caps and file-upload limits don’t apply to Codex (Codex help). Check again after the December change.

FAQ

Does this start on December 1? OpenAI only says “starting in December.” It hasn’t given a day.

Will my plan price change? OpenAI’s announcement says nothing about price changes.

Will I hit limits more often? OpenAI says it expects “most users won’t need to wait for their usage to reset more often than they do now.” Your own Settings → Usage history is the best way to check once the change is live.

Does Instant chat really not count? That’s what OpenAI’s Help Center says: “Instant chat will not draw from your usage.”

Does this affect the OpenAI API? No. The announcement covers ChatGPT plans. API usage, including Codex with an API key, is billed separately at API rates.

Does turning off model training change anything? No. Training settings and usage limits are separate. If you want to check those settings too, see how to stop ChatGPT and Codex training on your data.

More ChatGPT, Codex and Claude plan guides are on the 7minai homepage.