Would be very nice indeed if Cursor had a continually running spend meter in the bottom bar or chat window somewhere always visible. I thought there was a meter somewhere but I cant find it, so not in a great place 
Especially helpful in max mode. Just a simple value
$840
that on hover shows:
Today: $125, This Month: $840. 7d: $400, 28d: $1240
You know they have something like that just not per day, but showing your total balance.
You didn’t even vote for yourself.
lol thats crazy throwing money at problems; that makes people dumber
When you consider a software engineer is on average $500-$1000 a day, really, thats insignificant.
I think it’s hidden in max mode, I’ve never seen it.
your reply confirms: being able to just throw money at problems makes people dumber ….
you are blinded from the actual issue here: a single prompt which includes the full DB schema file in context will produce better quality design/architecture than a 10x back-and-forth requests which pulls in random partial files into context
I’m not so sure about that if I have a huge project (and I do) with thousands of code files, why wouldn’t I pay five dollars for it to find and fix the bug, and then I’ll just review it afterward to make sure it’s correct?
Why would that make me stupid?
What’s the point of me wasting a week on it and losing several thousand dollars?
I think I’m only now understanding what you mean keep in mind it could make things very slow, no?
But still, it’s nice as an option.
I’m waiting for the day when, for every chat message I send, I’ll see how many tokens it was, how much it cost me, etc.
I am saying putting those $5 in a single or two requests which uses the full 1M context of Gemini and includes exactly the full kernel files from the thousands files (my codebase is 700k hundred of files too) (without trying to second-guess what should be truncated) (and then some couple sliding-window-64k-tokens-contexts to fix remaining simple linter errors) will produce better design/architecture for the feature than letting Agent mode do 100x request pulling partial files into context. That is litterally the definition of DESIGN/ARCHITECTURE: You need the FULL VIEW, FULL CONTEXT AHEAD; and latest versions of Cursor is arbitrarily interfering with that, over-reaching its jurisdiction.
Wouldn’t have anything to do with speed, just show the current billing data which is already being collected and aggregated and billed by Cursor request by request.
Who is saying anything about what is happening? This is about cost, not how the models/LLMs are being used. Yes, more context, less requests theoretically should be better. That has nothing to do with this feature request, thats another thread.
Dollar is inflationary, therefore showing it makes no sense. Tokens is quantifiable and are a better measure of usage than cost. If you’re worried about costing, perhaps Cursor is not the tool to use as its only going to get more expensive for folks with little to no programming budget.