ChatGPT and Codex Will Share Usage Limits in December. Here's What Changes

OpenAI is changing how paid ChatGPT usage works starting in December 2026. Plus, Pro, and Business subscribers will share a usage allowance across ChatGPT and Codex as the company brings the task-performing capabilities of ChatGPT Work into the main ChatGPT experience.

According to OpenAI's announcement explaining the change, ordinary Instant chats will not draw from that allowance. The bigger question is how much research, coding, and other demanding work your subscription will cover. The announcement does not provide a complete numerical breakdown of the December allowances.

What Is Actually Changing?

ChatGPT Work and Codex already share an allowance. Work handles longer assignments, such as researching a topic or creating a spreadsheet, while Codex focuses on programming and technical work. December brings that shared usage arrangement into the main ChatGPT experience as Work's capabilities move there.

The practical implication is that covered activities draw from the same supply. For example: A lengthy research assignment could leave less capacity available for a coding project later. That is how a shared allowance works, but it does not establish that every subscriber will receive less usage.

What Are the Current Limits?

OpenAI's current Work and Codex pricing documentation provides estimated local-message ranges for Plus and Standard Business users. Examples include five to 45 messages with GPT-6 Astra, 15 to 160 with GPT-6.1 Sol, and 350 to 3,000 with GPT-6 Luna per five-hour window.

Those are estimates, not guaranteed message counts. Cloud tasks can consume more allowance, and weekly limits also apply. These figures describe the current arrangement, rather than a guaranteed December allocation.

Pro 100, Pro 200, and Pro 500 currently have no five-hour Work and Codex usage limit. They still have an included allowance and a weekly limit. Removing the five-hour restriction does not mean unlimited work.

Why Counting Messages Does Not Tell You Much

The upcoming shared allowance will be measured in tokens, the pieces of information a model processes and generates. More advanced models and complex assignments generally consume more.

Consider the difference between asking what a Windows error means and asking an assistant to examine logs, research possible causes, edit a script, and test the result. Both start with one message from you. One involves considerably more work.

Your prompt is only part of the equation. Files, conversation history, tool results, reasoning settings, and the generated answer can all affect consumption. A short request can launch a long assignment.

That makes message counts a pretty poor fuel gauge.

Where Do You Check Your Usage?

For current Work and Codex usage, open Settings > Usage to check your remaining allowance and reset times. Codex users can also enter /status in an active command-line session. OpenAI says the combined ChatGPT and Codex usage display will appear under Settings > Usage with the December transition.

Pay attention to which limit you have reached. On plans with both five-hour and weekly limits, you need capacity remaining in both. Having weekly usage left does not prevent you from exhausting the shorter window.

To get a practical feel for your own workload, check your remaining allowance before and after a demanding task. It will not predict every future assignment, but it gives you more useful information than counting how many questions you asked.

Who Benefits?

The clearest immediate winners are subscribers receiving the separate Pro allowance increases. OpenAI's updated Pro plan documentation says Pro 100 and Pro 500 now include twice their previous Work and Codex allowance.

Pro 200 subscribers who had the lower allowance receive an increase to match the higher allowance. Those already receiving the higher allowance keep it. OpenAI has also canceled the previously announced Pro 200 reduction. These updates apply automatically, without waiting for renewal.

Plus receives no increase through this particular announcement. A doubled allowance also does not guarantee twice as many completed assignments, since the work and settings still affect consumption.

People who mostly use ordinary Instant chats have less reason to worry about the shared meter because those conversations are excluded from it. That exemption does not make every other ChatGPT feature unlimited.

Who Might Find It Less Helpful?

People who regularly combine ChatGPT assignments with coding have the most reason to watch their usage. A shared allowance means those activities compete for the same available capacity and that capacity isn't exactly set in stone yet.

Users who need predictable output may also find the new system frustrating. It is difficult to plan around a subscription when you cannot reliably translate its allowance into finished reports, spreadsheets, or coding sessions.

However, calling December's change a universal downgrade would be premature. OpenAI has not supplied enough detail in the announcement to make a complete before-and-after comparison for every plan.

Should You Pay for More?

If you exhaust the upcoming allowance, OpenAI says your options will include waiting for a reset, upgrading, or buying credits for eligible features. The company expects most users will not have to wait for resets more frequently than they do now. That remains an expectation rather than a guarantee for your particular workload.

Before upgrading, find out what actually exhausts your allowance. A more expensive plan makes sense when it solves a recurring interruption, especially if the work saves you enough time to justify the price.

My concern is predictability. Most people do not want to learn token accounting to find out whether their subscription can finish a spreadsheet. Bringing more capabilities into ChatGPT could be convenient, but users still need a clear answer to a basic question: How much useful work am I getting for my money?

Geek Take

My take is that OpenAI wants to make ChatGPT feel like one product while getting a better handle on what each subscription costs to operate. Bringing research, file work, and coding closer together makes sense. Once those activities overlap, separate usage allowances become harder to explain. This would also explain the retirement of GPTs in favor of plugins.

There is also a significant difference between answering a question and sending an agent off to work. Ask for a definition, and it produces an answer. Ask it to investigate a problem, read files, and check its work, and that can keep the system busy for quite a while. I can see why OpenAI would want a meter on that to control usage.

But let's be honest, there is a sales angle here, too. When more activities draw from one allowance, heavy users have more reason to consider an upgrade or buy additional credits. OpenAI lists both as options when the allowance runs out. That does not prove the change is designed to squeeze subscribers, but pretending there is no business incentive would be a little disingenuous. They are in the business of selling credits.

Keeping ordinary Instant chats outside that allowance also makes sense to me. People can keep using ChatGPT for everyday questions, while the more expensive work gets metered.

I don't object to paying more for something that does more. I do object to having to guess what my money buys. If OpenAI combines the allowances and gives users enough capacity to keep working, fine. If the same monthly payment starts buying less useful work, a simpler interface will not make that feel like an improvement and i don't fell like I'm on an island with that opinion.

Jim McMahon Jim McMahon, aka Corporal Punishment, is the founder of MajorGeeks.com. He has spent decades testing software, troubleshooting Windows, and helping users cut through the nonsense. He loves real freeware, hates bloatware, and runs on caffeine, sarcasm, and questionable choices.

comments powered by Disqus