on day 2 of my 2 week Hobby tier, with price = free, and 2 weeks trial of pro, I had used up my 500 fast premium requests.
on day 7 i used slow premium requests, no problem.
on day 8, today, it says “You’ve reached your trial request limit”.
Steps to Reproduce
On my account, after time since creation >= 8 days, any use of premium modells gives the message " You’ve reached your trial request limit" in vscode/cursor editor.
Expected Behavior
it is supposed to let me use slow premium requests as lon as time since creation < 15 days.
Operating System
Windows 10/11
Current Cursor Version (Menu → About Cursor → Copy)
version: 0.43.6
vscode: 1.93.1
commit: not possible to copy/paste from about. You should have instructions here and a button to copy the entire about box and paste it here.
Yes I am getting the same error ‘You’ve reached your trial request limit. Please upgrade to Pro to continue.’ Even though my account still says ‘Pro Trial
8 days remaining’. Can I have help to fix this asap please?
The trial is two weeks and comes with 250 fast requests (half the 500 monthly fast requests in pro). After using up the 250 fast requests, you can still use other features of pro, but will need to upgrade for the 500 fast requests and slow requests.
ok, if that is the case, I highly recomend updating the information in pricing!
"Hobby
Free
Includes
Pro two-week trial
2000 completions
50 slow premium requests
It could say:
Hobby
Free
Includes
Pro two-week trial with 250 fast premium request
but only 2000 completions
and only 50 slow premium requests
Which is also not a good writeup, so I would say:
Hobby
Free
Includes
Pro two-week trial but limited to
** 250 fast premium requests
** 50 slow premium requests
Monthly 2000 completions
Monthly 50 slow premium requests
But is that even correct??? I’m now at least 29 days away from my account creation, so I’ll check back in a few days and see if 2000 completions reset and if slow premium will activate.
I wish there was a simple panel to toggle completions, slow/fast premium, and more explanation of what actually activates a premium request. Now I assume it has to do with which model I chose, so some modells are always premium while others are always not premium, but it was unclear when I started the demo
Note, this is the first time I have checked in after the pro trial ended. It then told me completions was only included in pro, so either i had to deactivate completions or sign up. Again it is not saying that hey your monthly 2000 completions will reset after a month from starting Hobby, and you may turn on complletions again then.
Actually, I’m not 100% certain what the completions are. Is it any code suggestions/predictions that are popping up as I type? Does that mean that if I decide to write out a line isteand of taking the suggested completion, I might end up using 5-10 completions while typing out a whileloop??? If so I need a very easy shortcut to disable completions.
I’m actually going pro for a few months, so all this is not really a problem, just some tips to you. As a pro user my biggest greevance is not having some continous view of my useage because the limits are very low seeing as I will actually use up all my fast requests in just 4-5 days.I also miss some sort of overview in the editor describing each model (including some icon or something indicating if the selected model is premium or not).
I now also tried claude 3.5, and it now says I have used my 50 monthly free claude 3.5. That means the 50 free do not come on top of the 250, they are only reset after the full first month. That is ok. But the “error message” when trying to use claude 3.5 says either upgrade to pro, or stay on hobby, where you get 200 “cursor-mini” requests.
What are cursor-mini requests?
And do I also get the promised 50 premium slow requests, which to me is 50 claude 3.5 requests.
I really like cursor ai, first ai tool to impress me, but the pricing list/description is really unclear and annoying and is the main cause of me not allready upgrading to pro
found even more info that is strange and confusing
from logged in to web account:
Usage (Last 30 days)
Premium models546 / 50
You’ve hit your limit of 50 fast requests
gpt-4o-mini or cursor-small197 / 200
You’ve used 197 requests out of your 200 monthly fast requests quota.
So this indicates we get 500 fast requests in the hobby->pro trial included for two weeks, but we also get 50 more.
the 50 more which are the monthly are here during the first month stated to be fast premium, but I expect that is a typo and that it will say 50 slow premium when I get in to the actuall regular first full Hobby month after the trial?
It also says 200 fast gpt-4o-mini OR 200 fast cursor-mini, which I assume are the not premium modells? This is not described at all in the Hobby tier limitations.
Another thing that sorely needs an update is the whack settings page, and specifically the entire API key UX needs a revamp - specifically - I should be able to put my API key for Anthropic and OpenAI alongside Cursor, without the warning, and with a F*CKTON more transperency.
When my limit is reached it should auto use my personal API key for whateve provider
Need to know when I am consuming my quota on tokens within Cursors prompting infra - or actually against the Claude API (their key or mine)
the ability to specificy which key to run against, same as droppong down the the model
the ability to have multiple Composers open and each window having its own key… see 3).
4.1) since I am using my own keys on the composers, I need to have a running tally of consumption. (ITS FREAKING AI - it should be able to count the fn consumption)
and a lot of other things that need to happen around Composer and Chat History - unacceptable that such little though has been put intto this by ALL providers:
OpenAI, Cluade, Cursor - atrocious design given that the primary interface is chat/composer to make it malevelently opaque for history etc should be shamed openly, loudly and oftern.
The fact that I spent a minute rebuilding the capability of SpecStory whereby I am able to directly yank cleaned chats and composer sessions in real-time into postgres, and have it searchable…,
and the fact that we dont have an open spec on the schema sitting in the vscDB file ■■■■■ – as we are paying to put that data in there , and we are relying on what is coming out - we should have complete control of that schema+db+data…
