Hi everyone,

I’m a ChatGPT Plus subscriber and I use ChatGPT professionally for long, complex legal-psychological analysis.

For this kind of work I almost always use High Thinking, because the additional reasoning quality is extremely valuable to me.

However, I can’t find clear information about how High Thinking usage limits actually work.

Could someone clarify the following?

• Are High Thinking limits based on messages, compute, or something else?
• Are there any published limits for ChatGPT Plus?
• Is there a way to see how much High Thinking usage remains?
• If the limit is reached, does it reset after a few hours, after a day, or after a week?
• Has anyone here actually reached the High Thinking limit while using ChatGPT professionally?

I’m not trying to reduce my usage—I actually want to rely on High Thinking for my work. I simply want to be able to plan large projects without worrying about unexpectedly losing access during an important analysis.

Thank you!

EricGT:

Please do not take this the wrong way, but the topic begins with one question, expands on it, and then ends with a somewhat different question—which I assume is the main issue you need addressed.

I can only infer what you are trying to accomplish, so I will explain how it relates to my own work and how I currently avoid a similar problem.

As a programmer, I use /goal for tasks that may run for several days without requiring my input. However, those runs were being interrupted by the dreaded message:

Selected model is at capacity. Please try a different model.

After several false starts, I now use a /goal prompt that instructs the process to pause itself between 11:00 UTC and 19:00 UTC. As I understand it, those tend to be peak usage hours for generative AI services such as ChatGPT.

I do not know exactly what part of the current prompt makes this work reliably, but it has been effective so far. Initially, I checked the run at each time boundary. When it did not pause or resume correctly, I used a Steer prompt to correct it.

Please ask if any part of that is unclear. I know it would have been confusing to me the first time I encountered it.

Hi Eric,

Thank you so much for your kind explanation. I think my question is actually much simpler.

What I’m really trying to understand is:

• Approximately how many High Thinking prompts are available for ChatGPT Plus users?
• If I reach that limit, how long do I have to wait before High Thinking becomes available again?
  • A few hours?
  • 24 hours?
  • A week?
• Is the reset based on a rolling time window or on a fixed daily/weekly limit?

I’m not asking about Codex or model capacity. I’m specifically asking about the High Thinking usage limit for ChatGPT Plus.

Thanks—that makes more sense to me now.

The only official information I could find is on the OpenAI Help Center:

https://help.openai.com/en/articles/20001354-gpt-56-in-chatgpt#reasoning-usage-and-fallbacks

It states:

ChatGPT displays when an allowance resets when that information is available.

The only place I know to check is under Settings → Usage in your ChatGPT account.

I am on a ChatGPT Pro plan, which has few, if any, usage limits, so I may not see the same information that appears for other plans.

Deleting my first reply which is of no longer of value in this topic.

Hi @constantia.degier,

Adding to the document @EricGT linked here his great suggestion on checking your usages, the current documentation and OpenAI does not specify whether the High reasoning limit for ChatGPT Plus is calculated by messages, compute or something else. It also does not publish an exact numerical limit, document a way to view the remaining allowance or state a fixed reset period.

The documentation only says that ChatGPT displays when the allowance resets when that information is available. If the reasoning limit is reached, ChatGPT may continue using GPT‑5.4 Thinking mini.

So unfortunately, there does not currently appear to be enough published information to calculate or plan your remaining High usage precisely. Whether anyone here has actually reached the limit would need to be answered from personal experience.

@EricGT I think the Usage and limits applies for Codex, if you meant this?