ChatGPT and Codex Will Share One Usage Pool in December
A shared pool sounds neutral. It isn’t. If you run Codex for hours and barely touch ChatGPT, you might come out ahead. If you split time evenly, the same pool can feel tighter than two separate buckets did.
Here is what is actually known as of November 14, 2026, what is still blank, and a short routine to prepare before the switch.
Jump to a Section
- What is changing in December 2026?
- Which plans are affected?
- What counts against the pool and what doesn’t?
- The facts still missing
- How do you measure your usage before December?
- Stay, upgrade or add credits?
What is changing in December 2026?
Starting in December 2026, OpenAI will give ChatGPT Plus, Pro and Business subscribers one shared usage allowance covering both ChatGPT and Codex, according to TechRepublic’s November 2, 2026 report citing OpenAI’s usage guidance. Until now, regular chat and coding-agent work were metered as separate things on most consumer plans.
Part of this already exists. OpenAI’s Codex help article says Codex, ChatGPT Work, ChatGPT for Excel and Workspace Agents use a shared allowance and credit pool on plans that use shared allowances or credits. The December change pulls more of ChatGPT into that same accounting.
When I checked the Codex help page, I noticed that it describes the pool in plain terms but gives no plan-by-plan numbers. It sends you to the pricing page for current limits instead.
Which plans are affected?
The December shared allowance applies to ChatGPT Plus, Pro and Business subscribers, per TechRepublic’s November 2026 coverage of OpenAI’s usage guidance. Free, Go, Enterprise and Edu accounts have no announced December change.
| Plan | December shared pool announced? | What to do now |
|---|---|---|
| Free / Go | No change announced | Nothing yet; Codex access stays limited |
| Plus | Yes | Log a week of Codex plus chat use |
| Pro (all tiers) | Yes | Note how often you hit limits today |
| Business | Yes | Ask the admin for per-seat usage history |
| Enterprise / Edu | No change announced | Follow your admin’s billing model |
Table built from TechRepublic (Nov 2, 2026) and OpenAI’s Codex help article, as of November 14, 2026. Plan names for Pro tiers vary; check your own billing page.

What counts against the pool and what doesn’t?
Instant chats will not count toward the shared ChatGPT and Codex allowance, per TechRepublic’s November 2, 2026 report. Heavier work does: Codex tasks and the advanced ChatGPT features that OpenAI folds into the pool.
Combined usage will be measured in tokens, and consumption depends on the model you pick and how hard the task is. OpenAI’s own help article says much the same: usage depends on the model, where the task runs, task complexity, context, reasoning, speed and tools. A long-running task can use substantially more than a short request.
That last point matters most. One sprawling agent run can cost more than fifty quick questions.
Two edge rules from OpenAI’s help page are worth knowing. Auto-review safety checks, when you are signed in with a ChatGPT account, are free and don’t count toward plan usage. And regular Chat Voice has separate limits and doesn’t consume Codex usage.
The facts still missing
As of November 14, 2026, OpenAI has not published the size of the December shared pool, the reset schedule, or the exact accounting rules for every advanced ChatGPT feature. TechRepublic says so directly, and I found that OpenAI’s public pricing page mentions only plan-level wording such as “Expanded Codex usage” for Plus.
Those gaps decide everything. A pool that matches your current combined allowance changes nothing for you. A smaller one changes a lot.
One more thing to keep apart: an October 29, 2026 update that doubled Work and Codex allowances for Pro 100 and Pro 500 and cancelled a planned cut for Pro 200 is a separate change. It does not tell you what the December pool will look like.
Don’t buy a bigger plan on a rumor. Wait for the numbers.
How do you measure your usage before December?
You measure it by logging one normal week of ChatGPT and Codex work and noting every time a limit notice appears. That gives you a baseline to compare against OpenAI’s December figures once they exist.
- Open your usage page. Check Settings, then Usage, in ChatGPT and screenshot what is shown today.
- Tag your sessions. For seven days, jot down whether each long task was Codex, Work or plain chat. A note on your phone is enough.
- Record every limit notice. Write down the day, the time and the reset time displayed. Limit hits are the number that matters.
- Separate API work. Codex used with an API key follows API billing, not the plan allowance, so leave it out of the tally.
- Count your seats (Business). Ask the admin for per-member usage so one heavy user doesn’t distort the picture.
Can you tell which tasks eat the most allowance?
Yes, roughly. Long agent runs with large context and high reasoning effort use the most, while short single-prompt tasks use the least. Keep a running list of your three heaviest tasks each week.
Stay, upgrade or add credits?
If you hit a limit under the shared pool, OpenAI’s help article lists the options: wait for the displayed reset time, upgrade, or, for eligible Plus and Pro users, buy credits without changing plans. Credits are the cheapest test of whether you actually need a bigger tier.
- You never hit a limit today: stay put and re-check one week after the switch.
- You hit limits a few times a month: stay, and price credits for those spikes.
- You hit limits weekly: wait for the published pool size, then compare tiers.
- Your whole team shares a Business workspace: pull per-seat history first.
Our earlier breakdown of ChatGPT Pro 200 Gets Half the Usage on October 30 shows how fast plan math can flip when allowances move. And if you also pay for Google’s side of the fence, GPT-5.6 Sol vs Gemini 4 Argon API Cost Before Promos End covers the API-cost angle.
Tips
- Screenshot your usage page this week, before any December rollout changes the layout.
- Run your biggest agent jobs once and note the before and after allowance.
- Keep quick questions in Instant mode; those stay outside the pool.
Warnings
- Don’t upgrade based on a forum post. Pool sizes were unpublished as of November 14, 2026.
- Don’t assume Free, Go, Enterprise or Edu follow the same schedule. No change is announced for them.
- Third-party blogs mix up the October allowance change and the December pool. Check the date on any figure.
FAQ
Will Instant chats use up my Codex allowance in December?
No. Instant chats are reported to stay outside the shared allowance. Heavier ChatGPT features and Codex tasks are the ones that draw from the pool.
Does the December change apply to Codex used with an API key?
No. Codex used with an API key follows separate API billing rather than the ChatGPT plan allowance, so it is not part of the plan pool.
Where will I see my combined remaining usage?
TechRepublic reports that combined remaining usage and reset times will appear under Settings, then Usage, in ChatGPT once the policy takes effect. The exact layout was not published as of November 14, 2026.
Sources
- TechRepublic: ChatGPT and Codex Will Share Usage Limits in December (Nov 2, 2026)
- OpenAI Help Center: Using Codex with your ChatGPT plan
- ChatGPT pricing page
