GitHub Moves Copilot to Usage-Based Billing

GitHub announced that, effective June 1, 2026, Copilot will move from a flat-rate subscription to usage-based billing. The $10 / month Pro and $19 / month Business plans stay the same, but each now includes a matching pool of “AI Credits.” Excess usage is billed per token.

Ars Technica

GitHub, owned by Microsoft, says it has “absorbed much of the escalating inference cost behind that usage,” yet the gap between a quick query and a multi-hour autonomous session made the old model “no longer sustainable.” Anthropic has made a comparable shift, charging Claude Enterprise users $20 per user plus variable usage, down from a $200 cap.

AI’s Free Ride Is Over and the Tab Is Big

Token rates vary by model: OpenAI’s GPT-5.4 Mini costs $4.50 per million output tokens, GPT-5.5 $30 per million. GitHub’s FAQ states charges “will be calculated based on token consumption, including input, output, and cached tokens, using the listed API rates for each model.” Simple completions and “Next Edit” stay free; code-review uses GitHub Actions minutes.

Ars Technica

GitHub says the new model aligns pricing with the value developers receive and gives them freedom to select models, a step it deems necessary to keep Copilot financially viable as AI infrastructure costs rise.

Ars Technica

Developers see the change as eroding value. A FAQ entry quoted a user: “This just wiped GitHub’s value moat — why should I stay?” Critics warn that long sessions and premium models could double or triple expenses for heavy users.

AI’s Free Ride Is Over and the Tab Is Big

Enterprise invoices now resemble utility bills; PYMNTS notes that for each dollar spent on AI models, firms spend $5-$10 on integration, compliance and monitoring. GitHub is adding admin caps at company, team or individual levels. Fredrik Filipsson of Redress Compliance predicts heavy Claude Enterprise users could see costs rise sharply.

AI’s Free Ride Is Over and the Tab Is Big

GitHub maintains base subscription fees are unchanged, while PYMNTS suggests heavy users may face substantially higher bills. No precise average-user cost projection is provided, leaving the overall impact uncertain.

AI’s Free Ride Is Over and the Tab Is Big

• “Today, a quick chat question and a multi-hour autonomous coding session can cost the user the same amount,” — GitHub

Ars Technica

• “ The company’s answer: usage-based billing aligns costs more closely to actual value and gives developers freedom to choose which models they use.” — GitHub

AI’s Free Ride Is Over and the Tab Is Big

• “Running powerful AI models at that scale costs money, and neither company is willing to offer unlimited access indefinitely.” — PYMNTS

AI’s Free Ride Is Over and the Tab Is Big

• “A single employee can generate thousands of AI interactions in a day.” — PYMNTS

AI’s Free Ride Is Over and the Tab Is Big

GitHub’s admin caps and Anthropic’s enterprise limits point to a broader shift toward utility-style pricing. CFOs at large firms are increasingly scrutinizing AI spend, suggesting usage-based models will become the norm as frontier AI infrastructure costs keep rising.

AI’s Free Ride Is Over and the Tab Is Big