The era of "all-you-can-eat" generative AI is ending. GitHub moved Copilot fully to usage-based billing on June 1, 2026, as the broader industry — including OpenAI, Anthropic, Google and Microsoft — shifts toward charging by token consumption.
June 1, 2026 · GitHub Copilot
GitHub Copilot Switches to Pay-Per-Token Billing
Flat-rate pricing is out. Starting June 1, 2026, Copilot charges by actual token consumption — heavy agentic users will pay more, while monthly plan prices stay the same.
$0.01
= 1 GitHub AI Credit, the new billing unit replacing Premium Request Units
$0
Code completion & Next Edit Suggestions stay unlimited
3
Token types billed: input, output & cache
Plan price = included Credits
Each column shows monthly cost; included Credits match the price 1:1. Exceed it, and you buy more.
June–Aug promo: +$30 Credits (Business) · +$70 Credits (Enterprise). Credits pool across an org; budget caps available.
Why the change
Copilot became an agentic platform — and offering quick questions and hours of autonomous coding at one flat price squeezed costs.
Flat rates were a land-grab during market acquisition. Now inference costs are visible, providers like Anthropic are also pairing seat fees with per-token charges. Transparent metered billing may become the industry standard.
Seen as fair
Price aligned with actual usage
Easier cost control via lightweight models & prompt tuning
Org-wide pooling and budget caps add flexibility
Concerns remain
Full-repo analysis & long sessions could spike bills
Tighter rate limits ahead of the transition
Large-scale agentic use is hard to predict
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…