GitHub Copilot's First Usage-Based Bill Just Landed, and Some Developers Owe the Price of a Used Car
2026-06-30
“The flat fee is dead, the meter is running, and a 29 dollar habit just turned into a 750 dollar surprise.”
June 30 marks the close of GitHub Copilot's first full token-metered billing cycle, and the invoices are landing like anvils. Developers are reporting jumps from 29 dollars to 750, and from 50 dollars to 3,000, with heavy agentic users seeing bills 10 to 50 times their old flat rate. The autocomplete that used to cost less than lunch now costs like a consultant.
The mechanics are tidy on paper and brutal in practice. Base prices held, Pro at 10, Pro Plus at 39, Business at 19, Enterprise at 39, but every plan now ships a credit allowance and then charges per token after that, input, output and cached, at published model rates, with no ceiling unless you go and set one yourself. The people getting torched are exactly the ones who bought the premium tiers for agentic refactors and multi-file changes, the workflows the marketing sold hardest.
Here is the cynical truth nobody on the pricing page will say out loud: unlimited was always a customer acquisition loss leader, and the bill was always coming. Charging by the token is arguably honest, the old model was unsustainable as agents got hungrier, but flipping the switch and letting users discover the meter at month end is a choice. The hard spending cap exists. They just did not turn it on for you.
- June 30, 2026 is the close of Copilot's first complete 30-day token-metered billing cycle since the model went live on June 1.
- Developers reported costs jumping from 29 dollars to 750 and from 50 to 3,000, with agentic users hit 10 to 50 times harder than the old flat rate.
- Base subscription prices were unchanged: Pro 10, Pro Plus 39, Business 19 per user, Enterprise 39 per user, each with a monthly credit allowance where one credit equals one cent.
- Every chat, agent-mode action, code review and CLI call draws from that balance at per-model rates, counting input, output and cached tokens, with no spending ceiling unless manually set.
- GitHub's own guidance urges users to enable a hard spending cap under Settings, Billing, before the next cycle begins.
- 01
Token billing is genuinely more honest about what AI coding costs, and the unlimited model really was unsustainable as agents grew greedier. Teams that set a hard cap, monitor usage and pick cheaper models for routine work can control the spend, and the shock is pushing the whole industry to think about efficiency rather than infinite free compute.
- 01
The power users, the exact developers GitHub spent years pushing toward agentic workflows and premium tiers, who opened a month-end invoice 10 to 50 times bigger than the flat fee they signed up for. Loyalty to the heaviest features turned out to be the most expensive position to hold.
No meter. No surprises.
The story isn't Copilot. It's the meter. Here is the calmer way to get your code reviewed.
- Flat price for the Full Suite. No usage meter, no month-end surprise.
- Free CLI: 90 roasts a month, no account needed.
- Privacy-first: a dry-run shows the exact payload, and your secrets never leave your machine.
Works in Claude Code, Cursor and Windsurf via MCP. Open source, and proud of it.
Got something the world should see roasted? Drop it.
A full teardown from €2,99. No mercy.