I would like to share my initial feedback regarding Cursor Pro. While I am satisfied with the tool’s features, I am very disappointed with how the usage limits are managed. Having started on January 10th, I already reached my quota by the 16th—after only six days of work. I deeply regret the total lack of visibility regarding real-time credit usage. For a PRO $20 monthly fee, I expected an uptime more aligned with professional use. As of today, I am seriously considering canceling my subscription.
Hey, thanks for the feedback.
If you used up the included (20) USD usage in 6 days, that means you’re using Agent pretty actively. For context, Pro includes (20) USD of API usage, but based on stats, Cursor daily Agent users usually spend around (60) to (100) USD per month. If you use Agent regularly plus premium models (Claude or GPT with Max Mode), the limit runs out fast.
On visibility, I agree the editor doesn’t have real-time tracking yet. For now, you can monitor usage here: https://cursor.com/dashboard?tab=usage (breakdown by model, tokens, dates). There’s also a community extension that tracks usage inside the IDE: Cursor Usage Tracker (Extension)
Options:
- On-demand usage: you can keep using Cursor after the limit, billed monthly at the same API rates
- Upgrade: Pro Plus ((70) USD included) or Ultra ((400) USD included) if your usage is consistently higher
- Optimization: switch from Auto to cheaper models, and avoid Max Mode unless you really need it
I would like to share my feedback following the cancellation of my Cursor Pro subscription. My intention is to provide a constructive perspective on quota management, which was the deciding factor for me.
While the tool is technically impressive, I encountered two major obstacles after only six days of work:
-
The “Pro” Designation: I exhausted my monthly quota in less than a week of intensive use. Given the $20/month price point, the current limits feel more aligned with a “Hobby” tier than a truly “Professional” one, where the tool is expected to be used for 8 to 10 hours a day.
-
Token Consumption through Error Iteration: This is my primary concern. The Cursor Agent often operates through trial and error—generating code, testing it, failing, and then retrying. Since tokens are consumed with every attempt, I feel that the user ends up paying for the AI’s generation errors rather than for the final result. Consequently, credits vanish rapidly during the tool’s internal debugging phases, rather than being spent on the actual delivery of requested features.
As a new user, I may be misinterpreting the underlying mechanics, but the lack of real-time visibility combined with this rapid consumption led me to rescind my subscription. I hope that, in the future, Cursor will offer more transparent real-time tracking or quotas that are better suited to daily professional workflows.
I wish the best of luck to those who continue to use this tool, which remains, in many other respects, highly promising.
1 Like