I hit context compact frequently, have only ever hit rate limit once and I am an avid user
I have frequently receiving below the scaring message recently. So I am wondering how to find out how many requests I have consumed and how many quota I have exactly.
1 Like
did they remove the forums from the top navigation menu on the home page or is it just me lol
3 Likes
Exactly, I don’t want to jump to conclusion but I think the backlash was too massive 
Can someone please tell me what is going on right now? Last thing I heard was Cursor changing their plans to move to a rate-limiting based approach, where you’d get limited temporarily after heavy use. I was abroad for 2 weeks and did 2 coding sessions this week, and I now have been locked for 3 days. Wasn’t the idea that you would be temporarily rate limited?
I really don’t understand anymore. I mainly use Claude Sonnet 4 and its thinking variant. I worked for hours the other day so I was not necessarily surprised that I hit the rate limit. What surprises me is that it seems to last until the end of the month?
The link to the forum has been deleted from the top of the website
9 Likes
So that the slaves don’t whine about the new prices)
They treat you differently, I use 180
I gotta drop this real quick but the fact that Cursor removed the Forum tab from their main navigation menu speaks VOLUMES. It tells you everything. WOW!
1 Like
I do appreciate @mdrain pointing the menu change out! I have reported it internally.
Most likely someone pointed out that docs are not shown in menu, not a sign of anything else.
@MarcoHuXam Cursor switched to a monthly included usage that is based on your consumption (context size as tokens and model cost). There are no more rate-limits.
The usage limit is until your subscription renews (so for 2 weeks now).
Following page could be helpful for understanding token usage and how to get more out of your plan.
It’s a very bad decision to cut back like that.
Now I tried to add $1 over the limit.
Half flew away ($0.5) for 1 promt where it was necessary to remove unused code and rewrite the duplicate code into a function
Today, I spent 12$ to attempt for the third time a simple 20 lines feature update.
The change is supposed to be mirrored in two separate but otherwise structurally identical files.
Claude just “forgets” to update both files but swears it does.
There is a A LOT of tool calls and failed tool calls.
I give it images, logs, database contents and it still doesn’t provide conceptually coherent changes.
The formatting and syntax is ok, but based on that it feels like everything is ok.
This is the first time I wasted hours with no progress, I just keep resetting the code and try again.
Using Claude 4 Sonnet Max Thinking, because the rest are worse and actually more expensive.
Why is it this bad?
My billing cycle is 29th of each month
Also, you could see the API Usage till the date
Here you could see the API usage for the Clude-4-Sonnet-Thinking is
Calculation
Sum input tokens for which usage is billed:
Total input tokens billed = Input (w/ Cache Write) + Input (w/o Cache Write)
= 3,275,318 + 1,075,555 = 4,350,873 tokens
Total output tokens billed = 224,892 tokens
Convert tokens to millions:
Input tokens: 4,350,873 / 1,000,000 = 4.350873 million
Output tokens: 224,892 / 1,000,000 = 0.224892 million
Compute cost:
Input cost: 4.350873 × $3 = $13.05
Output cost: 0.224892 × $15 = $3.37
Total estimated cost: $13.05 (input) + $3.37 (output) = $16.42 USD
But as you can see, I am seeing the $28.47 amount in here. Could you please explain this to me.
5 Likes
New month, new $20 dollars, but I only got 12 total rows in the detailed usage that were ‘included’, and it immediately started charging me additional requests. Can you please investigate.
Has anyone ever had a response from support? almost 2 months now, multiple emails, zero response. Happening for hundreds of others from what I can assess on the forum posts.
1 Like
jeeez… i reach my cap 6 days after my billing cycle, if I’m lucky, it keeps updating… 
kind of at a loss.. if i spend more than 100 dollars this month, i will have to go back to claude code.. which, i honestly dont like as much as cursor.. and even then now we are being capped at anthropic. It is really really frustrating to not be able to rely on pricing or capping anywhere. My company doesn’t pay for this.. I have too, and have really come to rely on using these tools.. waa, i know.
i just hit the usage limit with the ultra plan 3 days before the subscription renews.
if i get less api usage next month then i will cancel and change to claude code
got 1100 dollars worth of usage in my last billing period. 550 this time. if I were getting an output more aligned with Claude code, I would not care. but this uncertainty is ridiculous. yesterday I was projected for 8/12 but now it’s done and I have barely used it. what a bait and switch from these guys