The Pro Plan Limits Deception: Demanding an Explanation for a Service We Pay For But Cannot Use

Ideas Feedback
21

Also, when it comes to pricing:

  • an hour of Gemini costs about $7.5
  • Grok 4 is a bit more expensive, but smarter
  • Claude 4 is way pricier, even without Thinking

So yeah, you probably want to figure out what your costs and potential profit could look like.

  • If you’re making $30/h from a task, giving 30% of that to a machine that’ll do it for you isn’t the worst deal.
  • But if you’re making $4/h, then maybe don’t use a $40/h machine for it, right?
22

cancel, and use gemini code or claude code or other ai dev ide

23

So what exactly are we paying $20 or $60 for? The “auto” mode is just a free alternative to basic, free models! If I’m paying money, it means I’m choosing and expecting to use a serious, advanced model — not the “auto” mode, which is a free version meant for kids’ entertainment.

24

Cursor is even in violation of EU-Regulation 2005/29/EG at this point. This is gonna backfire so bad. What a waste.

25

lol calm bro

i know your pain, we also feel the same

have you experience with using API key? and see how fast you will burn $20

27

LOL, look what I found

28

Auto uses frontier models sometimes. Also auto is unlimited and free in background mode and only uses frontier models.

29

Pro should be hobby. a software of this caliber should not only be 20 dollars a month. if the usage rates were more in line with Claude Code but with the extra bells and whistles cursor provides (other models, IDE) I would gladly pay 500+ a month. my issue is if I pay 200/mo for Ultra and get to 28 out of the 30 days before being limited, why do the next 2 days cost me 150 LOL

30

For anything simple use auto. That might get you over the last two days.

31

I am not proficient enough in coding (my degree is MechE) to trust auto. it was really confused when I tried to use it last time

32

Try it again. Trust me you don’t need a degree in computer engineering to use auto. You just need practice. Just like using a calculator. The more prctice the better.

33

I canceled the Pro plan about two months because I could not understand the plan limits. I can see that the situation is getting worse.

36

image
image464×733 51.7 KB

There’s an extension called Cursor Usage that you can use on Chrome to see all your requests by provider which gives you an idea of where you’re at within the limits. It’s not perfect but it’s a good guidance I think.

Disclaimer: I’m the maker of the extension.

37

It’s incredible to have to install a third-party extension just to see how it’s used. It’s the basics

38

Unfortunately this extension is showing approximate info and not what Cursor is counting actually. Cursor does not track requests anymore, therefore all calculations shown there are applicable for average users only and not to a specific usage.

If one user has lower context usage (less files/rules/mcps… attached or context needed) they could get more than 225 of Sonnet 4, where on other side if another user has high context usage (lots of attachments, rules, MCPs,…) it leads to high consumption above average and less requests. This is why the Token breakdown is shown in Dashboard Usage

I wrote up a more detailed explanation of token usage and what you can do to optimize it.

40

Just pure disappointing. Cursor started to be annoying, and got the same usage message as others did.

Came onto the forum to see if it’s just me, same or even less usage than typical on my end. Surprise, wasn’t just me.

I paid annually by the way, from almost inception. Yes, stupid, but I wanted to support what they were trying to do from the early onset. There is no way I’ve even used the usage I have in the 7-8 months of usage.

I was expecting bugs, annoyances as they developed, but the absolute last I was expecting was a bait and switch on annually paid pro plan.

Subscription cancelled. With Claude Code and Gemini Cli, being more useful as well, ide integrated, and Cursor team did exactly the opposite when faced with new competition.

FYI - slow requests was a good solution.

41

I support it to the end. This is an indirect deception.

42

I cant stand the whole AI market atm, the whole pay per token ■■■■ need to go, especially when you spend half of them correcting the model.

There were a time when you paid for data, calls or texts on your phone before common sense took over.

43

I was not aware of this nuance, it’s really good to know, thanks for the info!

44

Im crazy w Cursor since they updated the Pro Plan and give new Ultra plan. Pro took me only 4 days to reach the limit of claude-4-sonet. The pricing of that model changed from 1x to 2x (But I think more). I have paid $20 for not only one week of Claude 4, and have to use free “AUTO” model? Nonsense.

I think cursor should remove plan $20 and focus on $200, like they want :slight_smile: