How to Check GitHub Copilot Premium Request Usage

GitHub Copilot's 2026 billing model introduced premium requests— a monthly allowance of high-capability AI completions and chat interactions that sits on top of the base plan. When you exhaust your premium request allowance, Copilot either downgrades responses to a lighter model or — on certain plans — bills overages per request. The tricky part: your premium request count is only visible inside GitHub's billing interface, and there is no way to see it alongside what you are burning on Claude Code, Codex CLI, or Gemini CLI. whoburnedmore closes that gap for the CLI-side of Copilot usage.

How do I check my GitHub Copilot premium requests?

GitHub exposes premium request consumption in two places. The billing settings page is the authoritative source for your monthly allowance and overage:

1. 1

   Open GitHub billing settings

   Go to github.com → Settings → Billing and Plans → Usage. Under the Copilot section you will see a usage bar showing premium requests consumed vs your monthly limit.
2. 2

   Read the current month's figure

   The number resets on your billing anniversary date, not on the calendar month. If you joined Copilot mid-month, your window may not align with the 1st.
3. 3

   Check for overage warnings

   If you are on Individual or Business plan without an overage budget set, you will see a warning when you approach the limit. On Pro+ plans with overage enabled, usage above the allowance bills at a per-request rate.

  Copilot Pro+ — June 2026 billing period  ─────────────────────────────────────────────  Premium requests used:     847 / 1,500  Resets:                    Jun 28 (14 days)  Standard completions:      unlimited   ↳ 56% of allowance used, 14 days remaining  ↳ at current pace: ~1,250 total (under limit)

What does whoburnedmore track for Copilot?

GitHub Copilot CLI — the terminal-facing agent component of the Copilot suite — writes local session logs in the same way Claude Code and Codex CLI do. whoburnedmore reads these to give you a token-level breakdown of Copilot CLI usage, which complements the premium request count on the billing page:

$ npx whoburnedmore↳ scanning ~/.config/github-copilot/sessions/…↳ found 234 entries across 17 sessions   GITHUB COPILOT CLI — last 30 days  ─────────────────────────────────────────────  2026-06-14   gpt-4o     841,200     claude-s.  312,900  2026-06-13   gpt-4o   1,204,700     claude-s.   87,400  2026-06-12   gpt-4o     674,300     claude-s.  441,600   ALL TOOLS — 30-day summary  claude code         18.3M     $61.40  codex cli            9.1M     $24.80  gemini cli           4.7M      $6.10  github copilot cli   5.8M     $14.20  ─────────────────────────────────────────  total               37.9M    $106.50

The cross-tool total is what makes whoburnedmore useful beyond what GitHub's billing page shows. Premium requests tell you about Copilot in isolation; this table tells you Copilot represents 15% of your combined AI coding spend — and whether that proportion makes sense for the value you get from it.

Premium requests vs raw tokens: what is the difference?

A premium request is GitHub's billing unit — it maps loosely to one multi-turn interaction with a high-capability model. The actual token count per request varies widely: a quick “explain this function” might be 800 tokens, while a multi-step refactor with tool calls might be 40,000 tokens, yet both count as one premium request. Token-level tracking (what whoburnedmore provides) gives you better insight into which types of interactions are expensive.

How is Copilot CLI cost estimated?

Copilot routes requests to different underlying models — primarily OpenAI's GPT-4o and Anthropic's Claude Sonnet, depending on the task. whoburnedmore applies each model's public API rate to compute an API-equivalent cost. On a flat Copilot plan you do not pay this directly, but knowing the API equivalent tells you whether the subscription is priced fairly for your usage level:

At mid-2026 API rates, a month of Copilot CLI usage that totals 5.8M tokens split roughly 70/30 between GPT-4o and Claude Sonnet comes to approximately $14–18 in API equivalent. If your Copilot Pro+ plan costs $19/month and you also get editor completions and priority chat, that math favors the subscription. If your actual CLI usage is lighter, it might not.

Which Copilot plan gives the most premium requests?

As of mid-2026 GitHub offers three tiers relevant to individual developers. The premium request allowance per month scales with the plan, but so does the price per overage request if you exceed it:

1. 1

   Copilot Individual — 300 premium requests/month

   Suitable for light chat and occasional agentic tasks. Heavy CLI use — tool-call loops on a large codebase — can exhaust this in a week or two.
2. 2

   Copilot Pro — 1,000 premium requests/month

   The tier most individual developers land on. Enough for daily use without constant worry, but a productive sprint week can still burn through 200–300 requests.
3. 3

   Copilot Pro+ — 1,500 premium requests/month

   Adds priority model access and the highest allowance. At this tier, overage charges rather than hard limits apply when you exceed the cap. 📊

What whoburnedmore does and does not cover for Copilot

whoburnedmore reads Copilot CLI local session logs. It does notconnect to your GitHub account or read your premium request count from the billing API — that number requires authentication with GitHub and lives on GitHub's servers, not your local machine. Think of whoburnedmore as covering the CLI-side token data, while GitHub billing covers the full Copilot surface (editor extension, chat, web).

Run npx whoburnedmore --dry-run to confirm what data is read and what would be submitted to the leaderboard. As always, no prompts, code, or file contents are ever read or sent — only numeric aggregates. 🛡️