Track Cursor AI Usage from Your Mac Menu Bar
Cursor Pro gives you 500 fast requests per month on GPT-4o and Claude Sonnet. AIUsageBar shows your running count and projected run-out date so you never hit the wall mid-sprint.
How Cursor's request limits work
- Cursor Pro includes 500 fast requests per month on premium models (GPT-4o, Claude Sonnet).
- After 500, requests fall back to slow routing on the same models — response times jump to 10–30s.
- The monthly counter resets on your billing anniversary, not the calendar month.
What AIUsageBar shows you
- Fast request count vs. your 500/month limit
- Days remaining until monthly reset
- Pace indicator: projected run-out date
- Slow request fallback usage
What is Cursor?
Cursor is an AI-first code editor built as a fork of VS Code. It integrates large language models — primarily GPT-4o and Claude Sonnet — directly into the editing experience via inline autocomplete, a chat panel, and Composer (an agentic mode for multi-file tasks). Cursor has become one of the most popular tools among developers who use AI heavily in their workflow, in part because it abstracts away model selection and offers a flat monthly pricing model rather than token-based billing.
That flat pricing comes with a catch: Cursor Pro gives you 500 "fast requests" per month. Once those are gone, Cursor falls back to slower responses on the same models, which many developers find noticeably disruptive to their flow.
Cursor's Request Limits: A Full Breakdown
Free tier
50 slow premium model requests, unlimited autocomplete using Cursor's smaller proprietary model (cursor-small). Primarily for evaluating the tool.
Pro ($20/month)
- 500 fast requests per month on premium models (GPT-4o, Claude Sonnet)
- Unlimited slow requests on those same premium models
- Unlimited autocomplete with cursor-small
Business ($40/user/month)
Unlimited fast requests, centralized team billing, org-level privacy controls.
Fast vs. Slow requests
"Fast" requests use a low-latency routing path — responses come back in under a second. "Slow" requests are queued on the same premium model but can take 10–30 seconds. The model quality is identical; the difference is entirely latency. For interactive chat this is tolerable; for autocomplete mid-keystroke it's not.
The reset date trap
Cursor's monthly counter resets on your billing anniversary, not the calendar month. If you subscribed on the 18th, your 500 requests reset on the 18th — not the 1st. Many developers discover this the hard way when they run out early in a month that felt like it should have reset already.
Why Tracking Cursor Usage Matters
500 fast requests sounds like a lot until you use Composer for a real project. A single Composer session that includes reading context, making edits, running checks, and iterating can consume 15–30 fast requests in one sitting. A week of intensive development can easily exhaust the monthly allowance.
The fallback to slow requests doesn't feel gradual — it's an immediate step-change in responsiveness. If you're in flow on a deadline, that shift to 10-second response times is exactly the kind of friction that breaks concentration.
Cursor itself shows you your remaining fast requests in settings, but not in a way that's visible while you're actively coding. By the time you notice you're running low, you may already be at 20 requests left with two weeks until the reset.
How AIUsageBar Tracks Cursor
AIUsageBar shows your Cursor fast request count and a countdown to your monthly reset directly in the menu bar. It also shows a pace indicator — based on your current burn rate, it projects whether you'll exhaust your monthly allowance before the reset date. If you're on track to run out in 5 days with 9 days to go, you'll know that now, not when the fallback kicks in.
Frequently Asked Questions
What counts as a "fast request" in Cursor?
Each completion or chat turn that uses low-latency routing on a premium model (GPT-4o or Claude Sonnet). A single Composer session with multiple back-and-forth exchanges can use 5–20+ fast requests depending on how many iterations are involved.
What happens when I run out of fast requests?
Cursor automatically falls back to slow requests on the same premium model. Response times increase from sub-second to 10–30 seconds. The model quality is the same, but the latency is significantly higher — which most developers find disruptive during active coding sessions.
When does my Cursor Pro limit reset?
On your billing anniversary, not the first of the calendar month. If you subscribed on March 18th, your 500 fast requests reset on April 18th — not April 1st. This catches many developers off guard.
Is there a way to buy more fast requests?
Not directly as an add-on for Pro. The only way to get unlimited fast requests is to upgrade to the Business tier at $40/user/month. There's no per-request top-up option for Pro subscribers.