ChatGPT and Codex will share one usage limit from December
OpenAI's help center now says that from December, Plus, Pro and Business plans get one usage allowance for ChatGPT and Codex, counted in tokens. Instant chat won't draw from it. OpenAI hasn't published how big the allowance is.
What changed
OpenAI's help article on usage limits now says that starting in December, Plus, Pro and Business plans share one usage allowance across ChatGPT and Codex, as the agent features of ChatGPT Work move into the main ChatGPT app.
Codex and ChatGPT Work already draw from one shared Codex limit. December's change pulls regular ChatGPT use into the same pool.
How it will be counted
| Question | From December |
|---|---|
| Unit | Tokens |
| What costs more | Advanced models and complex tasks |
| What costs nothing | Instant chat |
| Where to see it | Settings → Usage: combined remaining usage and reset times |
| At the limit | Wait for the reset, upgrade, or buy credits |
OpenAI's own forecast is that most users won't wait for a reset more often than they do now.
What OpenAI hasn't said
The article gives no allowance size for any plan and no exact date in December, and it doesn't mention Enterprise or Edu. Until those numbers appear, any "X tasks a day" figure you see quoted is a guess.
What it means for you
- You run long Codex jobs: from December your ChatGPT chats on advanced models come out of the same budget as your Codex jobs, so keep simple questions on Instant.
- You pay for both ChatGPT and a separate coding tool: watch your combined usage for a week in December, then compare Claude Code and Codex plans side by side before deciding whether one plan covers both.
Try it
Here's a task I'm about to give an AI agent. Before doing it, plan it so it uses as little work as possible. 1) List the files or inputs you actually need. Mark anything I pasted that you don't. 2) Split the task into the fewest steps that still let me check the result after each one. 3) For each step, say whether a fast model would do or it needs a reasoning model, and why. Don't start the task. Give me the plan only. The task: [describe the task] What I'll paste or point you at: [files, docs, data]
We've gone back and forth a lot in this chat. Write one brief I can paste into a fresh chat next time so I don't repeat all of this. Include: the goal, the facts and decisions we settled, the format I want back, and anything you got wrong earlier that I corrected. Leave out the dead ends. Keep it under 200 words, as a plain list.
Sources
- OpenAI Help Center, Understanding ChatGPT and Codex usage limits, checked November 2, 2026