OpenAI is merging ChatGPT Codex usage limits into one shared pool for Plus, Pro and Business plans, starting in December 2026. Instead of tracking coding and chat separately, you will draw from a single token-based allowance. OpenAI has not yet published the exact numbers.
Here is what is confirmed, what is still unknown, and how to avoid getting caught out when your coding sessions start eating into your chat time.
Quick facts:
- Plans affected: Plus, Pro and Business.
- Timing: “starting in December,” with no exact day given.
- Measured in tokens, not message counts.
- Plain instant chat is reported to stay outside the metered pool.
- Per-plan allowance sizes have not been published.
Skim that list and you have the gist. The rest of this post covers the details that matter for your wallet.
What exactly is changing with the shared usage allowance?
Until now, Codex, OpenAI’s coding agent, has had its own limits separate from regular ChatGPT. That is changing. According to explainx.ai’s review of OpenAI’s Help Center, Plus, Pro and Business usage “will be shared across ChatGPT and Codex.”
Heavier work will count more. OpenAI says advanced models and complex tasks use more of your allowance. That means a long agentic coding run could use up the same budget you would otherwise spend on image generation or deep research.
One caveat: Startup Fortune notes its own reporting leans on a checkout notice shown to some subscribers, not a standalone OpenAI announcement. Treat the December timing as reported, not guaranteed.
How will ChatGPT Codex usage limits work after the merge?
The table below compares the old setup with what has been reported for the new one.
| Area | Before | From December (reported) |
|---|---|---|
| Codex coding | Its own separate limits | Draws from the shared pool |
| Heavy chat tasks, agents | Counted separately from Codex | Same pool as Codex |
| Plain instant chat | Not metered | Still outside the pool |
| Where to check | Varies by feature | Settings, then Usage |
| At the limit | Wait for a reset | Wait, upgrade, or buy credits for eligible features |
The practical effect is simple. You get one number to watch instead of several, but a busy coding day can now leave less for everything else.
Why is OpenAI doing this now?
The merge fits a bigger plan. OpenAI’s Tibo Sottiaux said on a podcast in early October that the company wants to fold ChatGPT’s chat and work surfaces into one experience, as reported by Startup Fortune. One interface is easier to run with one meter behind it.
Codex limits have also been a moving target this year. The Help Center coverage lists several changes since July, including a removed and then returned five-hour window, and a promised reset after a September outage. A single pool is simpler to explain, even if it is not obviously more generous.
What should you do before December?
Nothing is required today, but a little prep helps. These steps keep surprises small:
- Note how much you lean on Codex versus chat in a normal week.
- Check the Usage page in Settings once it appears, and bookmark it.
- Move big, repetitive coding jobs to lighter models where that works for you.
- Decide now whether you would rather upgrade or buy credits if you hit the cap.
Doing this first means you will know within a week whether the new pool suits you. If you are on a higher tier, our look at how Pro usage limits doubled and the guide to claiming monthly API credits cover other ways to stretch your plan.
What is still unknown?
Quite a lot. OpenAI has not said how many tokens each plan gets. It is also unclear whether the five-hour window, the weekly cap, or both survive in the combined system, and which features credits will cover.
The Go, Free, Enterprise and Edu plans were not mentioned in the coverage, so their status is unknown. OpenAI says it expects most users will not wait for resets more often than they do now, but it has not shared data to back that up.
Frequently Asked Questions
Will my ChatGPT and Codex limits really be combined?
Yes, for Plus, Pro and Business, starting in December according to reports. You will have one token-based allowance across both.
Does normal chatting use up the shared pool?
Plain instant chat is reported to stay outside the metered allowance. Heavier tasks like agents, coding and complex requests are what draw it down.
Where can I see how much usage I have left?
OpenAI says combined remaining usage and reset times will appear in Settings, under Usage.
What happens if I run out?
You can wait for the reset, upgrade your plan, or buy credits for eligible features. Which features credits cover has not been spelled out yet.
Is the exact December date confirmed?
No. Coverage only says “December.” Watch the Usage page in your settings for the switch.
Our take
One pool is easier to understand, but it is only a win if the allowance is big enough. Until OpenAI publishes real numbers, light users will probably not notice, while heavy Codex users should expect to feel it. If coding with agents is your main use, hold off on any annual upgrade until you see the December numbers.
Leave a Reply