Please Keep ChatGPT and Codex Usage Limits Separate in December 2026

ChatGPT Feature requests
, ,
1

Please keep separate usage limits for ChatGPT and Codex.

ChatGPT is something we rely on every day, for all kinds of questions and tasks — both important and everyday ones. Users who spend a lot of time developing projects, working on ideas, experimenting, and learning through the browser should not have to sacrifice their ability to fully use Codex as a result.

ChatGPT and Codex serve different purposes, and their usage limits should remain separate.

P.S.
A little about why this is a problem from my perspective:

I personally believe that conversations with ChatGPT in the browser should be unlimited, or at least as close to unlimited as possible.

I completely understand why Codex needs usage limits. Coding agents perform complex, resource-intensive tasks, and reasonable limits make sense.

But regular conversations with ChatGPT are different.

ChatGPT still makes mistakes, misunderstands instructions, forgets context, and sometimes confidently provides incorrect information.

Users then have to spend additional messages explaining what went wrong, correcting the AI, and restoring context. Eventually, ChatGPT just says something like “You’re right, I apologize” or “Good catch!”

And now those corrections could potentially consume the same allowance we need for Codex?

That doesn’t seem fair.

I use ChatGPT to think through ideas, discuss projects, solve problems, and plan development before sending actual coding tasks to Codex. Sometimes these discussions take dozens or even hundreds of messages, especially when the AI makes mistakes.

I don’t want to have to choose between having a long conversation with ChatGPT and having enough Codex usage left to actually implement what we discussed.

I’m not asking for unlimited Codex. I understand why it needs restrictions.

What I’m asking for is unlimited, ChatGPT conversations in the browser, with Codex usage limits kept completely separate.

Please don’t turn every conversation, correction, or brainstorming session into a calculation of how much coding capacity we’re losing…

2

I support keeping ChatGPT and Codex usage limits separate. If OpenAI decides to combine them anyway, a real-time usage dashboard must be available before the change takes effect. It should show the remaining allowance, reset time, and consumption by product, model, and task. Without that transparency, I will consider the change a downgrade to my paid plan and reconsider renewing.

3

I agree with this.

I’m a paying ChatGPT Plus subscriber, and I’m deeply concerned about the proposed shared usage allowance for ChatGPT and Codex.

I use ChatGPT extensively for professional work, including document production, business management, research, and complex projects. To be clear, I’m referring primarily to Chat, not Work. I’ve deliberately avoided Work because, in my experience, its usage allowance is consumed so quickly that it becomes difficult to accomplish substantial tasks.

My concern is that extending a shared, token-based allowance to advanced Chat usage could introduce similar interruptions into workflows that many paying subscribers have come to depend on.

From a business perspective, I also question the wisdom of this decision.

At $20 per month, Plus is already a meaningful financial commitment for many subscribers, particularly in markets where purchasing power and exchange rates make dollar-denominated subscriptions expensive. The next tier, Pro, starts at $100 monthly—five times the cost of Plus—putting it beyond the bracket of many users.

This creates a potentially damaging gap: subscribers who rely heavily on ChatGPT for meaningful professional work may find Plus increasingly restrictive, while Pro remains financially inaccessible.

These are precisely the users who have found enough value in the product to integrate it deeply into their professional lives. Introducing restrictions that diminish that value risks pushing them towards competing AI platforms offering more accessible alternatives.

Usage limits should not undermine the practical value of a paid subscription, particularly for users whose primary interest is productive, sustained use of ChatGPT rather than Codex or autonomous Work features.

I would strongly encourage OpenAI to reconsider combining these allowances, preserve the existing Chat usage arrangements, or at least offer subscribers the option to retain separate allowances for ChatGPT and Codex.

The concern is not simply about usage limits. It is about preserving the value, affordability, and reliability that made ChatGPT worth paying for in the first place. Because if that is compromised, ChatGPT loses its competitive advantage

4

I like your dashboard idea!

5

Settings > Usage > Analytics
and
Settings > Profile
All the stats you just asked for.

6

The separate limits for chat and Codex are the product’s only competitive advantage. Essentially, you are asking users to switch to something else. I would be very sorry to leave, but it wouldn’t make sense to keep paying for a subscription with this limitation in place.

7

It doesn’t work for me. The dashboard suggestion he/she made would add real-time knowing use of tokens/limit per request made so we would have much more transparency.

The current profile/analytics show just estimates of unknown “chats” (?) (i use work though?) and I can’t tell what is eating away the credits/tokens unless I’m actively monitoring each request before and after we order the model, the dashboard real-time idea would be much more precise and useful in my opinion.

But I think we won’t be seeing this technology anytime soon, maybe because of added costs (?).

I still like the idea though. I hope it gets implemented eventually.

8

If we are to pay the same money for ChatGPT, Claude or Gemini with the same usage limits, why would I choose ChatGPT over Claude?

If they go ahead with this change, I will move my business elsewhere.

9

The distinction between ChatGPT and Codex usage is important, especially for Plus subscribers who regularly use both services.

ChatGPT is valuable for research, extended analytical discussions, problem-solving, and everyday assistance. Codex serves a separate development workflow. Combining their allowances effectively puts these activities in competition, even though subscribers currently have access to both.

The proposed real-time usage dashboard would certainly improve transparency, but it wouldn’t resolve the underlying concern. Knowing precisely how quickly an allowance is being depleted doesn’t compensate for potentially having less of it.

OpenAI’s assurance that most subscribers won’t encounter limits more frequently is also insufficient. The relevant question is whether existing subscribers, particularly those who use both products extensively, will retain equivalent access for the same subscription price.

There should be a clear, quantitative comparison of current and proposed allowances before this change takes effect, including consumption rates for advanced reasoning and Codex tasks.

More importantly, OpenAI should either preserve separate allowances or guarantee that the combined allowance provides at least equivalent practical capacity for subscribers who actively use both services.

Unifying the interface is a worthwhile improvement. Reducing flexibility or effective functionality shouldn’t be a condition of receiving it.

Until OpenAI provides those guarantees, this looks less like an improvement to the subscription and more like a potential reduction in its value.

10

OpenAI, this is an unacceptable way to treat paying subscribers.

You are asking customers to accept a fundamental change to their subscription’s practical value, without first publishing the numbers needed to judge its consequences.

You call it a unified allowance. But if using ChatGPT for research and complex reasoning now eats into the same budget needed for Codex, what exactly are we gaining?

If the combined allowance does not preserve our existing practical access, this is a price increase in everything but name.

And your reassurance that “most users” will not experience more frequent resets is meaningless without transparent, verifiable usage comparisons.

How many tokens will each plan receive? What are the model-specific costs? What happens to subscribers who regularly use both reasoning-intensive conversations and coding agents?

Why should customers trust a promise when OpenAI refuses to provide the figures necessary to evaluate it?

Exempting Instant chat is no consolation. We pay for capable reasoning models precisely because Instant cannot replace them in serious work.

Even worse, your official guidance suggests upgrading or buying additional credits after exhausting the shared allowance. That sends a terrible message: customers may have to pay extra to preserve the workflows their existing subscriptions previously supported.

I am tired of subscription services treating unilateral reductions in flexibility as progress.

Stop presenting a potentially worse subscription as a product improvement. Publish the numbers. Preserve separate allowances or offer an opt-out. Respect the customers who have supported your products.

If OpenAI proceeds without protecting existing access, I am prepared to cancel my subscription and take my money elsewhere.

Customer loyalty is earned. It is not an unlimited resource you can quietly spend.

11

I fully support this request to keep ChatGPT and Codex usage limits separate.

As a paying ChatGPT Plus subscriber, I use ChatGPT extensively as a technical and engineering assistant, often working on complex projects that require long, continuous conversations, advanced reasoning, troubleshooting, and iterative development.

I do not use Codex, and I see no reason why integrating Codex into a shared usage budget should negatively affect my existing ChatGPT experience.

My main concern is not the technical implementation of the new limits, but their practical impact on paying customers.

I need to be able to work with ChatGPT as I do today, without facing additional interruptions, unexpected usage limits, or forced waiting periods in the middle of an ongoing project.

I am also increasingly concerned about frequent changes to subscription conditions and usage policies. Customers who pay for a service deserve stability, predictability, and transparency.

I am not asking for more features, higher limits, or additional privileges. I simply expect the same level of service and continuity that I already pay for.

If the new policy results in more restrictive usage conditions than those I currently experience, I will cancel my ChatGPT Plus subscription and move to an alternative provider. I have no intention of paying more simply to recover functionality or usage capacity that I already have today.

I strongly encourage OpenAI to reconsider this decision, or at the very least provide a clear guarantee that existing Plus subscribers who do not use Codex will not experience any degradation in their normal ChatGPT usage.

Please do not make loyal paying customers bear the cost of integrating services they never asked to use.

12

I strongly support this request, particularly as a ChatGPT Plus subscriber who makes extensive use of BOTH ChatGPT High reasoning and Codex.

My situation illustrates a potential problem with the December 2026 changes.

I regularly exhaust my Codex usage allowance, including the five-hour limit, while independently continuing to use ChatGPT with High reasoning as my standard setting.

Under the current arrangement, exhausting my Codex allowance does not prevent me from continuing to use ChatGPT High.

If both services are moved to a shared token allowance, these two activities may start competing for the same resources.

Importantly, merging usage allowances does not necessarily mean reducing capacity. A sufficiently generous combined allowance could preserve the current experience.

However, OpenAI has not published enough quantitative information to establish whether that will actually be the case.

The statement that “most users” should not experience more frequent resets is not particularly reassuring for subscribers who already regularly reach their existing limits.

I believe we should collectively ask OpenAI to:

  1. Publish the exact Plus token allowance, consumption rules and reset periods before implementation.

  2. Explain whether exhausting Codex usage will prevent continued access to ChatGPT High.

  3. Demonstrate that existing subscribers can retain comparable combined usage capacity.

  4. Preserve separate allowances or guarantee minimum protected capacity if the shared system cannot provide equivalent access.

  5. Explain what protections or remedies will be offered to existing subscribers who experience a material reduction in functionality.

As an EU consumer, I also believe that any potentially material reduction in a paid digital service deserves clear advance notification and proper consideration of applicable consumer rights.

I am not claiming that a reduction has already been demonstrated. I am asking OpenAI to provide the figures necessary to determine whether one will occur.

I would encourage other subscribers who regularly exhaust Codex limits while also using High reasoning to share their experiences here. Collecting concrete usage patterns would make this discussion considerably more useful.

The objective should be measurable preservation of existing combined capacity, not merely greater transparency about potentially reduced limits.

13

OpenAI, back off. Do NOT force regular GPT-6 Medium/High chats into the shared Codex/Work token pool. We pay for advanced reasoning, not rationing every conversation. Preserve our dedicated chat allowances and publish the actual limits—or watch loyal Plus and Pro subscribers walk away.

14

I really don’t understand what OpenAI is thinking with this decision.

I’m a Plus subscriber, and I use ChatGPT and Codex almost every day. And I mean A LOT. Sometimes we spend hours analyzing documents, writing emails, discussing complicated situations, or working on my software projects.

And honestly, I love it. That’s exactly why this announcement worries me so much.

ChatGPT and Codex are two completely different things in my life. Why should spending hours coding mean I might not be able to use High reasoning later for something important?

But there’s something even bigger that has bothered me about AI from the beginning.

The dependency we’re developing on this technology is honestly frightening.

And I’m going to make a pretty brutal comparison here.

This is starting to remind me of the way heroin and fentanyl dealers exploit dependency. Not because AI is a drug, obviously, but because of the power that comes with making people dependent on something and then controlling their access to it.

Think about it. You give people something incredible. They start using it every day. They build their work, projects, routines, and even parts of their lives around it.

And then one day the rules change.

Want to keep using it the way you used to? Maybe pay more.

That’s what scares me. Not AI becoming smarter than humans, but humans becoming so dependent on AI that companies can dictate the terms whenever they want.

And before someone says “just upgrade to Pro” — not everyone can afford $100 or $200 a month. I’m already paying for Plus because this technology genuinely matters to me.

I understand computing power isn’t free. But please, keep ChatGPT and Codex separate, or at least guarantee independent allowances for both.

I don’t want to have to choose between working on my software projects and having a proper conversation with ChatGPT.

I really hope OpenAI listens to its users before making this change. I don’t want to leave. I just don’t want to see something this good ruined.

15

I fully support keeping separate usage limits for ChatGPT and Codex/Work. Realistically, being able to use Chat for brainstorming and basically everything else without usage impact is the only reason I still use ChatGPT over just using Claude. Without that versatility I see no reason not to just switch completely and save $20 a month.

16

I’m guessing this is about money. Because it doesn’t seem to serve the users.

Combining Codex limits with Chat thinking modes eliminates the largest benefit of using ChatGPT versus other services.

I’ll have to choose between design/planning/ documentation/recreation instead of doing them all between different agents.

I already maintain strict agent guidelines to keep my Codex usage minimal. I don’t need combined limits.

Chat already has limits.

Keeping them separate is important.

OpenAI is implementing a bottleneck with this choice, and I think it’s deliberate.

It will hurt my workflow and general use of the platform by forcing me to choose what is important.

17

I’m really not a fan of this direction, and I think the announcement glosses over some legitimate concerns for people who make substantial use of both ChatGPT and Codex.

I’m a fairly heavy Codex user, primarily working with Astra 6 and occasionally 6.1 Sol. I regularly approach or exhaust my weekly allowance during intensive development work. As of this morning, I was already at 67% of my weekly usage with four days remaining before reset. That’s not particularly unusual for me when I’m actively developing something.

I don’t necessarily have an issue with integrating ChatGPT and Work into a more cohesive experience. There’s some genuine utility there. What I take issue with is the decision to consolidate usage allowances for workflows that serve very different purposes.

I use ChatGPT for research, technical discussions, architectural planning, and general problem-solving. Codex is where I spend considerable time actually developing software. Having those activities compete for the same finite pool of resources introduces an unnecessary constraint that doesn’t exist in quite the same way today.

The announcement makes a point of doubling the existing allowance, which sounds generous on paper, but doubling a resource allocation while simultaneously expanding the number of things consuming it doesn’t inherently translate into more usable capacity. Without transparent model-specific consumption rates and a meaningful comparison against the existing limits, there’s no real way for subscribers to evaluate whether this is an improvement or simply a repackaging of what they’re already paying for.

I also find the suggestion that most users won’t encounter more frequent limits somewhat unconvincing as reassurance. That’s probably true for casual users, but those of us regularly pushing the existing limits are precisely the people most likely to feel the consequences of this change.

I’d much rather see OpenAI integrate the experiences while retaining independent usage allowances, or at least give subscribers the option to keep them separate.

Improving the product experience shouldn’t have to come at the expense of predictability in how we use the resources we’re paying for. And frankly, I’d prefer to spend my time developing software rather than strategizing about which model or conversation is going to eat into my remaining weekly budget.

18

Dear OpenAI Team,

I am writing to express my profound disappointment and strong opposition to the recently announced changes to the ChatGPT Pro plan, particularly the decision to consolidate ChatGPT and Codex usage into a single shared quota.

I am a ChatGPT Pro subscriber, and I deliberately chose your platform over alternatives such as Claude. This was not a casual decision. One of the primary reasons was the ability to engage in extensive and frequent conversations through ChatGPT Web, using advanced reasoning models without constantly worrying about usage limits.

My workflow relies on two tools that serve clearly distinct purposes:

  • ChatGPT Web: I use it extensively for research, discovery, technical planning, architectural analysis, problem-solving, decision-making, and everyday questions. It is my primary environment for thinking, exploring ideas, and doing intellectual work.

  • Codex: I use it for software implementation and development, turning the decisions and plans previously developed in ChatGPT into working code and tangible results.

This complementary relationship is precisely one of the greatest strengths I see in the OpenAI ecosystem.

The decision to merge these two usage quotas directly threatens this workflow.

Under the new model, every complex conversation in ChatGPT could consume resources that I need for software development with Codex, while every implementation task in Codex could reduce my ability to research, plan, or reason through problems in ChatGPT afterward.

This introduces usage concerns that previously did not dictate how I worked. Instead of freely leveraging the capabilities of both tools, I will now have to monitor, manage, and potentially restrict how I use them.

I find it particularly questionable that this change is being presented as an improvement through the announcement of doubled Work and Codex usage allowances, while simultaneously introducing a shared quota that could significantly disadvantage customers who rely heavily on both tools.

I also do not believe that offering unlimited access to the Instant model addresses the issue. I did not subscribe to ChatGPT Pro merely to receive quick answers. I subscribed specifically to use advanced reasoning models for complex tasks. Replacing that level of reasoning with Instant is not an equivalent alternative.

I understand that OpenAI needs to manage computational resources and develop sustainable business models. However, I believe that shifting this pressure onto customers by forcing previously independent tools to compete for the same usage allowance is a misguided decision.

Furthermore, I am concerned about the lack of transparency regarding the actual impact of this change. Announcing a unified quota without clearly explaining how it compares to the previous limits, or how much each type of activity will consume, makes it difficult to determine whether customers are receiving a genuine improvement or an effective reduction in service.

I therefore formally request that OpenAI:

  1. Reconsider the consolidation of usage quotas for existing ChatGPT Pro subscribers.

  2. Preserve independent usage allowances for ChatGPT Web and Codex, maintaining the complementary relationship that originally motivated customers like me to subscribe.

  3. Publish concrete, verifiable comparisons between the previous and new usage limits, particularly for customers who extensively use advanced reasoning models.

  4. Explain what protections or guarantees existing subscribers will receive to ensure that this transition does not diminish the capabilities they originally paid for.

  5. Consider grandfathering provisions or other protective measures for existing subscribers if the new conditions result in a material reduction in service.

I want to be absolutely clear: My complaint is not about OpenAI charging for its services. My complaint is that I am willing to pay for a premium service, but I am not willing to accept the deterioration of its essential capabilities under the guise of an improvement.

The freedom to use ChatGPT Web intensively without compromising my ability to work with Codex was one of the determining factors in choosing OpenAI over competing platforms.

If that advantage disappears, so does a fundamental part of my motivation to maintain this subscription.

I am seriously considering canceling my ChatGPT Pro subscription and evaluating alternatives such as Claude. This is something I would genuinely prefer to avoid, as I value OpenAI’s technology and the ecosystem you have built.

I hope this feedback will be treated as a genuine concern from a paying customer who uses your products extensively and wishes to continue doing so, but who strongly disagrees with the direction of this commercial policy.

I would appreciate a specific and substantive response to these concerns, rather than a generic explanation about increased capacity, flexibility, or usage allowances.

I also explicitly request that this complaint be reviewed by a human representative and escalated to the teams responsible for ChatGPT Pro product strategy, pricing, and usage policies. Please confirm whether this escalation has taken place and provide a support case reference if available.

Sincerely,

A deeply disappointed ChatGPT Pro subscriber.

20

I fully support this request.

I’m also a ChatGPT Plus subscriber, and I recently started a separate discussion expressing similar concerns, particularly about preserving dedicated usage for GPT-6 High conversations.

I primarily use ChatGPT for learning, extended discussions, and creative projects, while only occasionally using Codex or Work. However, I completely agree that the concern goes both ways: people who rely heavily on Codex shouldn’t have their development workflows restricted by unrelated ChatGPT usage either.

These are different tools serving different purposes, and subscribers should be able to benefit from both without having their existing usage capacity reduced.

I hope OpenAI will listen to the growing number of users expressing these concerns and reconsider the decision.

I’ve also shared my perspective here:

Thank you for speaking up about this!