GitHub CopilotPro
Month · premium requests16% left
Premium requests253 of 300 used
Headroom · for Copilot people
Your Copilot premium requests, counted.
Copilot gives you a fixed number of premium requests each month, and every model spends them at a different rate. Headroom keeps what's left in the Windows taskbar, with a countdown to your reset.
Windows 10 & 11 · signed auto-updates · no account needed
How Copilot's premium requests actually work
One monthly allowance, a multiplier per model, and a reset that ignores your signup date.
Copilot stopped being one undifferentiated pool of "requests" when premium requests arrived. Your plan now includes a fixed number of them per month: 50 on Free, 300 on Pro, 1,500 on Pro+, 300 per user on Business and 1,000 per user on Enterprise. Everything expensive draws from that number. Code completions and the included base model are the exception on paid plans: they stay unlimited, which is why running the premium pool dry does not leave you staring at a dead editor.
What makes the allowance hard to reason about is the multiplier. A premium request is not a prompt; it is a unit of billing, and each model spends a different number of them per send. The included base model costs nothing on paid plans, while frontier models cost a multiple of one request every time you hit enter. Two developers with identical habits and identical plans can end the month hundreds of requests apart purely on model choice.
Running out is undramatic on a paid plan and abrupt on Free. Paid plans fall back to the included base model, and if you have switched on additional paid usage, requests past the allowance bill as overage rather than stopping. Free has no such floor: the premium pool and the separate completions and chat allowances simply end until the reset.
Premium requests
The number your plan refills each month: 50 on Free, 300 on Pro, 1,500 on Pro+, 300 per user on Business, 1,000 per user on Enterprise. Agent work, frontier models and premium chat all draw from it. Unused requests do not roll over into the next month.
Model multipliers
Each model spends premium requests at its own rate, so the pool drains at the speed of the model you picked rather than the number of prompts you sent. The included base model is free of charge on paid plans; the expensive models are exactly the ones you reach for when the work is hard.
Completions & chat allowances
Copilot Free meters two more things that paid plans leave unlimited: monthly code completions and chat messages. Headroom shows whichever allowances GitHub actually reports for your account, so a Free plan gets all three meters and a paid plan gets the one that can run out.
What Headroom shows for Copilot
Read from your own GitHub sign-in: the same numbers your editor shows, and an honest error when GitHub changes something.
What's left, next to the clock
The tray icon shows your premium requests as percent left, straight from the allowance and remaining count GitHub reports. The flyout adds the raw numbers (253 of 300 used) because a percentage is easier to glance at and a count is easier to plan with.
Pace against an even burn
A tick marks where an even burn through the month would put you. Because multipliers make the drain rate depend on model choice rather than effort, "burning fast" on the 12th is the signal to drop back to the included model while the allowance still has somewhere to go.
Overage, stated plainly
If you have enabled additional paid usage, requests past the allowance keep working and start costing money. Headroom shows the overage count as its own line rather than pinning the meter at zero and pretending nothing is happening.
Fair questions about Copilot
When do GitHub Copilot premium requests reset?
Monthly, and for individual plans the allowance refills on the 1st rather than on your subscription anniversary. GitHub reports the exact reset instant on your account, and that is the date Headroom counts down to, not a calendar guess. Unused premium requests do not roll over.
How many premium requests do I get?
Copilot Free includes 50 a month, Pro 300, and Pro+ 1,500. Business seats include 300 per user and Enterprise 1,000 per user. On paid plans the included base model stays unlimited, so running the premium pool dry does not leave you without Copilot.
Why is one request costing me more than one premium request?
Every model carries a multiplier, so a single send can draw several premium requests from the pool. The included base model bills at zero on paid plans, while frontier models cost a multiple of one request each. The multiplier, not the number of prompts, decides how fast the allowance drains.
What happens when I run out of premium requests?
Premium models stop until the reset. On paid plans the included base model keeps working, so Copilot does not go dark. If you have enabled additional paid usage, requests past the allowance bill as overage instead of stopping, and Headroom shows that overage count as its own line.
How do I check how many Copilot premium requests I have left?
GitHub's own answer is the usage page in your account settings, and your editor's Copilot menu repeats it. Headroom reads the same numbers your editor reads and keeps them in the Windows taskbar, with a countdown to the reset, so checking costs a glance instead of a tab.
Copilot is rarely the only plan you're burning. Headroom also watches: