Since June 18, 2025, GitHub Copilot meters every premium request against a fixed monthly allowance. Enter your plan and usage to see when you'll run out — and what the same work would cost elsewhere. Free, no signup, runs entirely in your browser.
1. Pick your plan
Pro
$10/mo
300 premium requests
Pro+
$39/mo
1,500 premium requests
A "premium request" = one agentic action, chat turn, or code-review run. Plain autocomplete is usually free. Not sure? Most heavy users land around 30–80/day.
Advanced: tune the Claude API cost assumption
Used only for the Claude API comparison below. Defaults assume a context-heavy agentic request. Claude Sonnet 4.6 pricing: $3 / million input tokens, $15 / million output tokens.
Your projected burn
Projected month-end usage
–
% of plan budget
–
Credits run out on
–
Projected overage cost
–
What this usage costs you, by tool
GitHub Copilot — your plan + projected overage
–
Cursor Pro (flat)
–
Claude API direct (Sonnet 4.6, pay-as-you-go)
–
Estimates only. Copilot overage billed at $0.04/premium request beyond your plan's included credits. Cursor Pro is a flat $20/mo with its own generous request quota. Claude API is usage-based — the figure uses your token assumptions above. Real costs vary with how you work.
Get alerted before you hit 80%
This calculator is a rough estimate. I'm building a dashboard that connects to GitHub, tracks your real credit burn live, and emails you before you run out — potentially with a breakdown of which features (chat vs. autocomplete vs. agentic) burn the most. Leave your email to get early access.
No spam, just a heads-up when the live dashboard is ready. Unsubscribe anytime.
How Copilot's premium-request billing actually works
Each plan includes a fixed monthly premium-request allowance. Pro includes 300/month and Pro+ includes 1,500/month; once it's gone, each extra premium request bills as overage at $0.04. Metered billing of premium requests began June 18, 2025.
Not every request costs the same. Each model carries a "premium request multiplier" based on its cost — for example, Copilot code review counts as 13× a normal request — so heavier models and agentic features drain the allowance far faster than plain autocomplete, which doesn't count at all.
Running out doesn't lock you out. You can keep using an included base model for the rest of the month; only premium-model and agentic work pauses, unless you've set a budget for paid overage.
GitHub's native usage page shows a running total but no daily rate and no "at this pace you'll run out on the 19th" projection — which is exactly when people get surprised.
The math this tool does: (premium requests used so far ÷ days elapsed) × days in month, vs. your allowance. Over 100% means you're on track to run out early. It's a rough estimate — because of multipliers, your real burn depends on which models you lean on.
Allowances, multipliers and the $0.04 overage rate are GitHub's published figures — see GitHub's premium-request billing docs. Figures current as of June 2026; GitHub can change them.