ChatGPT and Codex Shared Usage Limit: What Changes in December for Plus Users in India
If you pay for ChatGPT and also dabble in Codex, the way your plan is counted is about to change. OpenAI has quietly updated its Help Center to say that from December, the ChatGPT and Codex shared usage limit becomes one combined allowance for Plus, Pro and Business users, measured in tokens instead of separate buckets. No big launch event, no keynote. Just a help article. And yet it touches every paid user who does more than casual chatting.
Here is what is actually changing, what it means for someone paying ₹1,999 a month in India, and how to avoid running dry halfway through a busy week.
What the ChatGPT and Codex shared usage limit actually means
According to OpenAI’s help page “Understanding ChatGPT and Codex usage limits”, the change arrives as the company folds the agentic side of ChatGPT Work into the main ChatGPT app. Once that happens, ChatGPT and Codex draw from the same pool.
Three details matter most:
- It is counted in tokens. Tokens are the small chunks of text a model reads and writes. A short question costs little. A long research task with files, or a coding agent working through a big repository, costs a lot more.
- Bigger models eat more. OpenAI says plainly that more advanced models and complex tasks use more of the allowance.
- Instant chat is outside the pool. Quick, everyday conversations on the Instant setting will not touch your shared budget.
You will see the combined balance and reset times under Settings, then Usage. If you hit the ceiling, OpenAI lists three options: wait for the reset, upgrade, or buy credits.
What OpenAI has not said yet is just as important. There is no exact December date, and there are no published token numbers per plan. So anyone telling you precisely how many tasks you will get is guessing.
Who should care, and who can relax
Honestly, a large chunk of Plus users can stop reading here. If your ChatGPT life is asking for recipe swaps, drafting emails and fixing a caption, you mostly live in Instant chat, and that is excluded. OpenAI also says it expects most people won’t hit a reset more often than they do today.
The people who should pay attention are the mixed users. Think of the freelance developer in Pune who runs Codex on client work in the morning and then uses ChatGPT Work to build a pitch deck in the evening. Until now those felt like two separate tanks. From December, it is one tank, and a heavy morning in Codex leaves less for the evening.
Content creators who use Work for long, multi-step jobs (research, spreadsheets, slide decks) belong in the same group. So do small agencies on Business seats where one person tends to be the “AI person” for the whole office.
How the limits already work in Work and Codex
The new pool builds on a system Work and Codex users already know. OpenAI’s separate help article on managing usage in Work and Codex explains that, depending on the plan, there can be a five-hour window and a weekly window, and you need room in both to keep going. Hit the five-hour cap and you wait, even with weekly allowance left.
The same page gives rough ranges for Plus. With the top GPT-6 Astra model, OpenAI estimates roughly 5 to 45 local messages in a five-hour window. Drop to GPT-6 Luna and the range jumps to 350 to 3,000. Those are OpenAI’s own estimates, not promises, and they show how wildly your mileage depends on the model you pick.
One more line from that page is worth remembering: switching models does not restore allowance in a shared pool. You cannot dodge a limit by hopping to a different model once you have spent the budget.
What it costs in India right now
On OpenAI’s pricing page as shown in India, ChatGPT Go is ₹399 a month, Plus is ₹1,999 a month and Pro starts at ₹10,699 a month. The iPhone App Store listing also shows a yearly Plus option at ₹19,900. Go is the cheap entry point, but the Codex and Work allowance is tied to Plus and above, so Go users are not part of this change.
Is Plus still worth ₹1,999 under one pooled limit? For most people, yes, because the pool mainly bites at the heavy end. If you are a working developer who already hits Codex limits every week, it may be time to price out Pro against simply buying credits for your busiest days. Do the maths on your own usage screen first rather than upgrading on panic.
Five ways to stretch your shared allowance
- Default to Instant for small stuff. Quick questions, rewrites and translations do not need a reasoning-heavy model, and Instant does not count against the pool.
- Pick the model for the job. Save the top model for genuinely hard problems. A lighter model handles boilerplate code and routine edits just fine.
- Keep context lean. Long threads with huge files pasted in burn tokens on every turn. Start a fresh chat when the topic changes.
- Check Settings, then Usage, on Monday morning. Knowing your weekly reset time lets you plan the heavy jobs.
- Batch agent tasks. One well-written, detailed request to Codex or Work usually costs less than five vague ones followed by corrections.
Our take
On paper, one pool is simpler. You stop guessing which bucket a task comes from, and you can pour your whole allowance into whichever tool you need that week. That flexibility is a real plus for people who bounce between writing and code.
The catch is transparency. A token budget with no published size asks users to trust a progress bar. Until OpenAI shares actual numbers per plan, Indian subscribers paying ₹1,999 a month deserve to keep a close eye on that Usage page in the first weeks of December, and to say so loudly if the “most users won’t notice” promise turns out to be optimistic.
If you are curious how other AI tools are pricing and rolling out here, read our breakdown of Pika H3 Flex for Indian creators and our check on whether Meta Muse is available in India yet.
FAQ
When does the ChatGPT and Codex shared usage limit start?
OpenAI’s Help Center says “starting in December” but has not given an exact date. Watch Settings, then Usage, in your account for the switch.
Does the change affect free or ChatGPT Go users in India?
OpenAI’s note names Plus, Pro and Business plans. Free and Go users are not mentioned, and Go (₹399 a month) does not include the Codex and Work allowance in the first place.
What happens if I run out before my reset?
OpenAI says you can wait for the reset, upgrade your plan, or buy credits for eligible features. Instant chat keeps working because it does not draw from the shared pool.
