What plan are you on? I never saw things like 0.3 credits for sonnet - not on the new neither on the old plan. What is happening here?
@proteus-dev I clearly stated that my screenshot is Usage Based Pricing, that can be on any plan. Max is counted in millions tokens processed in, out, cache write and cache read). All the cost is detailed in the usage stats on mouse over.
From my experience this seems to have been a tool call.
I am sorry that‘s so far away from my own experience of using cursor that i fail to grasp it. A single thread you say? People use opus max for huge big brain tasks. If its a single thread, the only valid conclusion here is, that barely 0 compute had been used.
Even a single sonnet no thinking prompt usually takes ~2 minutes to complete if you are doing serious work.
OP had their cost included in the plan.
We have very different usage patterns.
My tasks may run for 30minutes or longer. And I do not use Opus as its not necessary to burn more ‘limits’ on it.
I was confused my usage-based pricing is FAR away of what you are shwoing. Not only is yours a fraction of what i got charged, it is also displayed in credits, which also did not happen in my case.
Yes I used Usage Based Pricing (over the included amount/rate) as its cheaper for me to continue with the task instead of switching model.
OPs charge was within the plan limits = not charged.
A single tool-call. Exactly. 0 compute had been used, no serious work done. Whats your point here?
Yea ok, but back to topic: you said he used these 15 in a single thread. I say , it must be nearly 0 compute and no serious work. Do you atleast understand where my problem is with taking OP‘s post serious?
You posted inaccurate assumptions about another users MAX requests. Therefore I used my own requests where tokens were used and charged. A tool call is not charged but the AI model needs to send or receive tokens for 99% of tool call. So the statement that there is ‘no serious work done’ is inaccurate.
I still fail to see how your examples of your experience connect with what OP shows.
OP and I used MAX, that connects perfectly and explains the number of lines shown in stats. there are 2 differences though:
- OP used Opus and I used Sonnet (both thinking)
- I used usage based pricing as mine was overage over the included rate, and OP had it within his plan.
As those are not really differences but just different models used and overages instead of waiting. If I were within limits it would also show as cost -.
Not sure what the issue is with OPs post anymore 
Hello forum.
Is anyone else experiencing momentary halts/ freezes then using Cursor? Last a few seconds but too frequent to not be frustrating. Have tried clearing my /AppData/Cache, dis/enabling some accelerated hardware features that I’ve read Cursor takes advantage of, but the problem persists.
Hi @erissatallan and welcome to Cursor Forum. Kindly create a separate Bug report with such issues, since it does not contribute to this topic directly. Also in a separate bug report with more details its more likely that the cause can be found.
I have tested it, and the maximum is 200 tool calls. You cannot break through this
1 Like
Just don’t put the C drive. Soft connect to the solid-state drive, it should be caused by the garbage principle of the Windows system
I have Cursor Pro ×2 ($384/year, renews Sept 17, 2025) — how does that translate under the new pricing structure? Do I keep both seats? And where can I see how much time/credit I have left with the new system?
it translates for work life balance - you can code for 1 hour after which you hit rate limit which is great, you will have more time for wife, kids or yoga practice 
id like to know too, cant seem to find an answer either