Yesterday, I spent the entire day testing GPT-5 intensively across different use cases, trying all available modes, including Max Mode and Non-Max Mode. This morning, I wanted to continue my work but was immediately met with this message:
You are projected to reach your GPT-5 usage limits by 8/9/2025 based on your current usage. Consider switching to a different model such as O3, asking for smaller changes, or enabling pay-as-you-go to avoid interruptions until your cycle resets on 9/7/2025.
This means:
I can no longer test GPT-5 intensively, even though this is supposed to be a promotional week.
The limit seems to be reached very quickly, yet there is no clear information on what the actual limit is.
The reset date is September 7, 2025, so I would have to wait nearly a month or pay extra.
Why are there such restrictive limits during a promo phase? Why aren’t the exact usage caps made transparent to the user? Is there any way to reset the counter or increase the limit without switching plans immediately?
Honestly, I’m quite disappointed that I can’t properly evaluate the model and get a full picture of its capabilities before hitting a wall.
While we are offering a generous free trial of GPT-5, we do have limits in place to be ensure people do not abuse the free period. The usage allowance is substantial, providing ample opportunity for users to test GPT-5 at no cost.
Once the free usage period is over, the model will return to API pricing as with all the others in Cursor.
I understand that, but in my view it would be fairer to clearly communicate the limit in advance, both in the promo announcements and directly in the interface. This way, users would have a sense from the start of how much usage is realistically possible before restrictions apply.
Especially with a new model that consumes significantly more tokens per request, this has an even greater impact on usage. A transparent display of current consumption in relation to the limit would help users better assess and plan their usage, particularly when testing the features intensively to get a complete picture.
Is there any way to reset the limit, or perhaps use a different variant of GPT-5 to continue testing? Also, does the usage limit apply per project or account-wide, similar to the standard usage limits?
This isn’t good, because I’m using the most expensive and fastest one, since I thought it was free for a certain amount of time, not for a certain amount of tokens.
Do I need to quickly switch to the cheaper one?
What’s the limit?
How much is it in credits?
But this is very important to me. Should I try to use the cheaper ones?
Because until now, I’ve mostly gone for the more expensive and faster one.
If it’s limited, I’ll work with the slower one at least.
I also don’t know what feedback to give it works fast, it’s smarter than any of the models I’ve used so far, it does amazing things, and so far it hasn’t messed up at all.
What other feedback are you looking for about the model?
Oh, and it’s really good at calling tools.
For most tasks, especially for creating new things, the base model is excellent.
But there are a few things for which I absolutely need the heaviest model:
Hi,
sorry to be rude, but why reply at all, if you do not like to provide the answer?
This is a recurrent theme of your answers.
Maybe law/politics are better option for you. We are serious people and I am not gonna accept these kind of responses which is just meant to save your time and waste ours.
Guys come on surely you have learnt from your issues around the pricing changes of last month. TRANSPARENCY is very important here. I was under the same impression, using it under the faster mode, and then i got the message pop saying hi your usage limit for gpt5 is going to run out on the 12th of August, is that just the allotment of the free week, or is that the entire lot for my account for this subscription period? If we had been told up front how you are going to manage this im sure many of us would have been happy to work on the slower version, but get twice as much usage (as the fast costs twice as much).
I get it you don’t want to disclose details, but at the same time you tie us as developers and users to a very specific, limit based usage model. We need details….
@cocode sorry that I have not met your standard. I will pass the info to the team. You are still able to test various GPT-5 models, just not unlimited. As others posted in various threads they receive an advance notification when their usage of GPT-5 would run out.
@Zaz thats meant for the GPT-5 free testing period.
@LARAMediaGroup we will announce once the model switches from free to regular pricing.
