The ChatGPT Codex usage limits your team works with today are about to change. OpenAI says that starting in December, Plus, Pro and Business plans will share one usage allowance across ChatGPT and Codex. That means a long coding session can leave less room for research, writing, and agent tasks in ChatGPT, and the other way around.

For a software agency, this is a budgeting question, not just a product update. Many teams pay for a handful of seats and use them for both client code and client documents.

I am Nitin from the Debugged Pro team. Below I stick to what OpenAI has actually published and what reporters have confirmed. Then I share how we would plan for it if you run delivery teams in India or the USA.

What OpenAI Announced About ChatGPT Codex Usage Limits

The change is spelled out in an OpenAI Help Center article on ChatGPT and Codex usage limits. It says the shared allowance is coming "as we bring the agentic capabilities of Work into the main ChatGPT experience." Work is the ChatGPT mode that runs longer, multi-step tasks on its own.

Here are the confirmed points, in plain words:

Who it covers: Plus, Pro and Business plans.
When: December 2026. No exact date yet.
How it is measured: tokens, the units of text a model reads and writes.
What drives cost: the model you pick and the task you give it. Bigger models and complex tasks use more.
What is exempt: Instant chat will not draw from your usage.
Where to track it: combined remaining usage and reset times will show under Settings, then Usage.
At the limit: wait for a reset, upgrade, or buy credits for eligible features.

OpenAI also says it expects most users will not need to wait for a reset more often than they do now. That is a reassuring line. It is also hard to check, because no allowance numbers have been published yet.

ChatGPT Codex usage limits diagram showing ChatGPT, Work and Codex drawing from one shared token allowance — ChatGPT, Work and Codex will share one allowance on Plus, Pro and Business plans from December. Instant chat stays outside it. Graphic: Debugged Pro original, based on the OpenAI Help Center.

How the news came out

Users spotted it first. TokenPost reported on October 30, 2026 that a notice on the ChatGPT web checkout page said ChatGPT and Codex would share a plan usage allowance from December. That notice gave no date and no allocation details.

The Help Center page then filled in more detail, including the token-based meter and the Instant chat exemption. Reddit threads about the change picked up hundreds of upvotes within hours, according to VGTimes. Many users asked how usage will be counted.

What OpenAI has not said yet

The exact December start date.
How big the shared allowance is on each plan.
How common requests convert into tokens.
How Free, Go, Enterprise and Edu accounts are handled under this specific change.

Plan around those gaps. Do not build a budget on numbers that do not exist yet.

How the Current Limits Work Today

Some of the structure is already live, because Work and Codex share an allowance now. OpenAI's guide to managing usage in Work and Codex explains it.

Five-hour and weekly windows

Depending on your plan, two limits can apply at once:

A five-hour window. A new window starts with your first Work or Codex message after the last one ends.
A weekly limit. This caps total included work across the week.

You need room in both to keep going. So a developer can hit the five-hour cap early, even with weekly usage left. OpenAI says Pro 100, Pro 200 and Pro 500 currently have no five-hour limit in Work and Codex, though an included allowance still applies.

Rough message ranges for Plus and Business Standard

OpenAI gives estimated local message ranges per five-hour window. It stresses these are estimates, not guarantees.

Model	Plus	Business Standard
GPT-6 Astra	5 to 45	5 to 45
GPT-6.1 Sol	15 to 160	15 to 160
GPT-6 Sol	15 to 150	15 to 150
GPT-6 Luna	350 to 3,000	350 to 3,000

The spread is huge. Model choice matters more than anything else here. Cloud tasks may also use more allowance than local messages.

The recent Pro and Business Premium boost

On October 29, 2026, OpenAI's Business release notes said Pro 100 and Pro 500 plans now include twice their previous Work and Codex allowance. OpenAI also canceled a planned Pro 200 reduction and brought those subscribers up to the same higher limit. Business Premium seats got more Work and Codex usage too.

So the direction is not only tighter. Heavy users on higher tiers just got more room. The open question is what happens on Plus and Business Standard once chat joins the pool.

Why ChatGPT Codex Usage Limits Matter for Agencies

Most agencies we talk to use ChatGPT in two very different ways. Developers run Codex for refactors, tests, and code review. Project managers, writers, and analysts use ChatGPT for specs, proposals, and research.

Often these people share the same plan type, and sometimes the same habits. A shared pool changes the math in a few ways:

Heavy coders feel it first. One long Codex task can use far more than a short request, per OpenAI.
Research in Work now competes with code. A deep research task and a refactor draw from the same bucket.
Model defaults become a cost lever. Running everything on the biggest model will drain the allowance fastest.
Deadline weeks get riskier. Hitting a five-hour cap during a client release is a real delivery risk.

A quick example from a typical team

Picture a 10-person agency in Pune serving a client in Texas. Four developers use Codex daily. Three PMs use ChatGPT for status reports and user stories, often in Work mode.

Today, the PMs' regular chat use sits apart from the Codex budget. From December, their non-Instant chat and Work tasks will share the same allowance with code work on each seat. If a PM and a developer share an account, which some small teams still do, they will now compete for the same tokens. That is one more reason to give each person their own seat.

If your team runs Codex in the cloud, our guide to Codex cloud environments for agencies covers setup choices that also affect usage.

A 7-Step Plan to Prepare Your Team

Seven step checklist for agencies preparing for ChatGPT Codex usage limits in December — A simple prep checklist for India and USA agency teams. Graphic: Debugged Pro original.

Audit who uses what. List every seat, its plan, and whether the person mainly codes, writes, or researches.
Check Settings, then Usage, weekly. Once combined usage shows up, record it. Real data beats guesses.
Set model defaults by task. Use lighter models for routine edits and small questions. Save the heaviest model for hard problems.
Push quick questions to Instant chat. OpenAI says it does not draw from the allowance. Train non-coders to use it for simple lookups.
Split big Codex tasks. Smaller, well-scoped tasks are easier to review and less likely to burn a five-hour window in one go.
Match seats to workload. Business Premium seats include more Work and Codex usage than Standard. Give them to your heaviest builders first.
Write a limit playbook. Decide in advance who can buy credits, when to wait for a reset, and when to move a task to the API.

When the API makes more sense

OpenAI says API usage is billed separately at API rates. For predictable, repeatable jobs like batch code checks or CI pipelines, the API can be easier to budget. You pay per use instead of guessing how much of a shared plan you have left.

Keep interactive, human-in-the-loop work in ChatGPT and Codex. Move automated, high-volume work to the API. That split also keeps your seats free for the people who need them.

ChatGPT Codex Usage Limits and Client Budgets

Agencies in India and the USA often bill AI tool costs as overhead. A shared, token-based pool makes that harder to forecast, at least for the first month or two.

Here is how we suggest handling it with clients:

Do not promise fixed AI hours. The allowance is not a fixed number of messages or tasks.
Add a small buffer in December quotes. Cover possible credit purchases during the switch.
Track usage per project where you can. ChatGPT Desktop shows per-chat credit usage on some credit-based plans.
Keep security rules the same. A budget squeeze is not a reason to paste client secrets into the wrong tool.

On that last point, our post on Codex Security for India and USA agency teams covers safe review habits that still apply.

Comparison of ChatGPT and Codex usage before and after the December shared allowance change — Before and after the December change, based on OpenAI Help Center articles. Graphic: Debugged Pro original.

What We Are Watching Next

Three things will decide how much this matters for your team:

The real allowance numbers for Plus, Pro and Business seats.
The start date, so you can avoid switching during a big release.
How Instant chat is defined in practice, since it is the only exempt use.

We will update our own internal playbook as OpenAI publishes those details. If you build your plan now on audits and model defaults, you will be ready whatever the numbers turn out to be.

Key Takeaways

From December, Plus, Pro and Business plans share one ChatGPT and Codex allowance.
Usage is measured in tokens and depends on model and task. Instant chat is exempt.
OpenAI has not shared the exact date or allowance sizes yet.
Five-hour and weekly windows already apply to Work and Codex on many plans.
Agencies should audit seats, set model defaults, and write a limit playbook now.

Need help planning AI tooling, seats, and workflows for a delivery team? Our software development consulting team works with agencies across India and the USA on exactly this.

About the Author

Nitin Kumar is the Lead Project Coordinator at Debugged Pro. He plans delivery for software and AI projects across India and the USA, and helps teams pick tools that fit real budgets and deadlines.