Today is June 30 — the close of the first full billing cycle since GitHub Copilot switched to usage-based billing on June 1. The community reaction has been immediate and loud.
The numbers making the rounds
On Reddit, X, and GitHub’s own discussion forum, developers are sharing screenshots of projected monthly costs. From $29 to $750. From $50 to $3,000. One GitHub discussion thread collected over 400 comments and nearly 900 downvotes. A developer post consisting of just two words — “Goodbye, Copilot” — was shared thousands of times.
GitHub’s Chief Product Officer Mario Rodriguez didn’t soften the message: “Copilot is not the same product it was a year ago.”
How the billing works
One GitHub AI Credit equals $0.01. Code completions and Next Edit suggestions remain free. The meter runs on agentic sessions, premium frontier model access, multi-step autonomous tasks, and code review. A single agentic session can cost $30 to $40. When credits run out, premium features stop — no more fallback to a cheaper model.
What this means structurally
This is the first major crack in the “unlimited AI for a flat fee” model that has defined developer tools since ChatGPT launched in 2022. GitHub led the way — Cursor switched in 2025, Windsurf followed in March 2026.
The structural reality: the marginal cost of AI inference doesn’t approach zero the way traditional software does. Every token has real GPU costs. And flat-rate subscriptions that don’t scale with usage lose money on every power user.
For developers, this accelerates migration to alternatives. Claude Code doesn’t meter tokens on its Pro subscription. And that’s probably doing wonders for Anthropic’s growth right about now.
Sources: GitHub Blog, GitHub Discussion, Visual Studio Magazine