I said $100M+ deliberately, wasn’t trying to inflate numbers without receipts.
Yeah, I missed the latest $500M ARR update, but that only proves my point more clearly : this isn’t some accident or “bug.” It’s calculated growth built on usage-based monetization.
You don’t scale to half a billion ARR by leaving token efficiency to chance.
only solution is to use deepseek model right now
Another is to use claude code
To me it looks like they are over-inflating Cache Token usage. Both Models were given a task document to follow on a blank slate/empty project. I’d also love to know how Gemini had 6m input tokens for a 1300 line document when claude only used 363k. So it’s either a bug or …
now I am back to using auto mode it consume 0 tokens most times and some times will consume reasonable tokens and is not bad
For me auto model is unusable. It asks me for confirmation for everything which makes it unusable for giving him some more advanced tasks.
that’s just a skill issue. You need more practice prompting and managing auto’s context. it uses the same models. it does use claude for some prompts. and there are ways you can trick it into using claude more consistently.
is this a joke?
Was this a complex refactor? I would love to know prompt and output. ironically I find CC always handles these better.
Anysphere is making a lot of money, no doubt about it. You have to have a viable project, or be an experience contractor to be satisfied with Cursor.
It’s good to see the bar being set a lot higher, there is no need for everyone and their dog to clutter up the global codesabse with ■■■■ code, and dilute the quality of code that is subsequently used to train LLM’s - I see too many bad habits LLM’s are picking up already
The current pricing says that Pro+ should have 675 sonnet 4 requests, I ran less than 70 requests before hitting my limit. I switch to Ultra, and after half a day of using it, the API cost was $35! It looks like Cursor is charging direct API fees, rather than by number of prompts like they usually do, where each continue button starts a new prompt. Clearly we are not getting the 675 sonnet 4 requests I was getting a couple weeks ago, with pro+ before hitting the limit.
The issue seems to be with the cache read being out of control, generally cache reads are supposed to use only 10% of the API costs, now that are 70x the token size of the input/output tokens themselves. This needs to be fixed.
I’ve literally used Cursor every day for the past year +, developed 12+ applications with it. I’ve used the pro, then metered usage pricing up until last month (for 7+ months with metered usage at same rate of use it would be $60/month in total usage every month this year). I made 10 minor updates to my webapp today, that I’ve worked on with cursor for 5+ months without this issue. With this issue, with the same amount of use, I’m getting charged $35 for a half a days usage! Look at my image, it literally processed 63 MILLION TOKENS! in half day when my Input tokens was only 564,901 and output token 264,911 This is a big issue.
They won’t admit it, no matter how much I say. Change the CC
No the community mod said it’s not a bug therefore it’s not a bug and we just don’t know how to use Cursor.
dan posted an update from the team in the other thread:
