one question claude say they give 30-40 request every 5 hours for 20$ but based on the input tokens, so is not really like 10 request (if you are using a big project) every 5 hours? trae is not for every country and github, windsurf, bolt, lovable no one is giving you a really “unlimited” or high claude 4 sonnet usage why? so we now cant generate code with claude 4 sonnet?
wth,so now you’re nerfing the new pricing plan even more? wtf? I’m so confused. You guys really just need to leave the software industry, or stop listening to VCs in your ear encouraging you to implement anti consumer business practices. Its obvious you hate your products, your customers and the only thing the cursor team cares about are its investors and their wallets.
2 Likes
Same, there’s actual no way to characterize this as anything other than fraud and extremely anti consumer.
It wasn’t a “bad call” They chose to commit fraud.
1 Like
When do we get to see our usage? When do we get some kind of indication as to how much of our current allotment we’ve used?
I’m paying $40 a month (previously had upgraded the pro plan when that was still on offer)
How can I see what I’m getting for this?
Hey! Hope i can shed some more light on this
Is there a time limit (e.g., it lasts for 3 or 5 days after the $20 credit is used)?
Yes, once the monthly limit is exhausted you’ll have access to the local limits that are reset every 5-24 hours on a best-effort basis (meaning we wan’t to keep you going, but can’t guarantee when it will happen). You can read more here: Cursor – Models & Pricing
Does it function like the old ‘slow requests’ system, where resting for a while restores some limited usage?
You could say that with the local limits as outlined above. However, if local limits are met you can either enable usage based pricing, upgrade tier or wait until next local limit (or monthly) resets
Happy to clarify where I can!
Is this meant to be a clarification of existing terms, or a complete overhaul of the pricing structure?
This is meant to clarify the recent changes we made!
What about the previously unaddressed rate limits, burst limits, and local limits? And how do the Pro+ and Ultra tiers fit into this new scheme?
Burst limits was a unclear terminology, it’s just a monthly usage limit you get with the plan which is spent on tokens at LLM providers for API pricing. With Pro it’s $20, Pro+ $60 and Ultra $200. The local limits are on top of this and varies, but they are proportional to the plan. E.g Ultra has 10x local limits to Pro
We missed the mark and we’re sorry about this. The underlying mechanics of the plan hasn’t changed. Using “unlimited” was an inaccurate term to represent it as it only covered Auto
Didn’t you guys say it was the “local” rate limit being used first, then the “burst” if you go over? What you’re describing is the opposite, the “burst” being the 20$ and the “local” being the “best effort” one.
2 Likes
We want the old Cursor AI back — with 500 requests and an unlimited pool, as well as a single-query agent with unlimited tool usage.
Not this latest exploitative change. We do not agree to pay 10 times more for API access
3 Likes
We want “generous limit” as Cursor team described at first place.
In pro plan. In all models. Especially Claude 4 Sonnet.
2 Likes
For me it’s not just the lack of proper communcation, and I never enabled API billing so (hopefully??) I was never charged more than $20.
It’s that I invested $20 per month on a product that can stop functioning while I work. The model with “slow” requests worked reasonably, and I would have expected that rate limiting would do the same: if I’m too eager, the editor slows me down a bit. But the editor can just stop working entirely (I won’t use AUTO mode on my codebase). Given that github copilot does not do this anymore, that’s the winning product for me at the moment. Cursor simply does not have an affordable plan that does the job I expect it to do.
1 Like
What do you mean? On “unlimited” until July 3rd my usage probably would cost $200+ using today’s crazy pricing. Today, just 4hrs of coding and it says I used up $15.
Why can’t you admit the (unlimited) was a mistake, it even felt too-good-to-be-true because I went from spending $60-$100 on usage based, and suddenly all my usage based is free? and now today suddenly my usage-based is 5x more expensive?
And I upgrade to Pro+ and I cannot revert to the original 500 + usage pricing model.
This is the wrong way to run a business. what are you guys smoking over there???
2 Likes
Seriously, the 500 + unlimited slow was just fine. Sometimes you switch to usage-based for extra fast, but it was fair pricing. Today I can’t even do my normal development on the $20/m pro. Had to upgrade to $60/m and within 4 hours of coding I’m already suddenly at $15/$60 of spend (25% completed)
Like all I’ve done is a few UI bug fixes and changes. This is insane
4 Likes
Yeah, I am cancelling my subscription. Gemini is going bonkers and gets stuck in loops, Claude rate limit hits after 4 prompts. Wild.
5 Likes
u got rate limit after 4 prompts in claude code?
The new pricing literally translates to “pro users can’t use claude sonnet 4”, be it max or not max mode.
Because once your monthly usage is capped, forget the grace period…it doesn’t do anything.
You get billed immediately… For Claude lovers, I think this is a call to use the main provider instead, since it doesn’t seem like cursor is in it for the community but for their pockets(I mean ideally its a business..hhaha)
Because just today on just three tasks I am down 15 dollars. The worst is that on one of the tasks, I got 3 bad results that I had to revoke, and I was already charged for that…hahah… I mean, I understand that it’s no one’s fault, but yeah, just come straight and tell us Cursor is now for exclusive, high-paying users.
1 Like
you’re naive to think you’ll get an answer
What is going on with this pricing guys. I am genuinely confused.
I know the discount for Sonnet 4 stopped a while ago but is it really true that I pay 0.80$ - 1.20$ per request??
It also says something crazy like: 1,616,971 in token, whereas when running gemini they range anywhere from 10k tokens to max 374,078
My credit card will be burning up quickly 
Yes, that’s true — and the reason for such prices is explained here.