Enterprise & Industry

GitHub Copilot shifts to usage-based pricing June 1 - why that's no surprise

GitHub Copilot's new AI Credit system means usage caps and potential price hikes.

Deep Dive

GitHub announced a major shift in its Copilot pricing model, moving to usage-based billing with AI Credits starting June 1, 2026. The new system replaces the current premium request unit (PRU) model with token-based consumption, including input, output, and cached tokens at published API rates. Base subscription prices remain unchanged—$10/month for Copilot Pro and $39/month for Pro+—but these now include monthly AI Credits matching their dollar value. Code completions and Next Edit suggestions will remain free, but agentic coding sessions that run long, multi-step tasks will consume credits. GitHub claims the change is necessary to maintain service reliability as Copilot evolves into an agentic platform with significantly higher compute and inference demands.

This pricing overhaul was foreshadowed when GitHub recently blocked new Copilot subscriptions and restricted model access for individual plans, including dropping Opus models entirely. Users who exhaust their credit allotment will lose access to the service, a stark departure from the current unlimited usage model. While GitHub frames this as aligning costs with actual usage, many users are already expressing frustration over potential price increases. The move reflects a broader industry trend where AI service providers are moving away from flat-rate pricing to consumption-based models as inference costs continue to rise. GitHub will preview the new billing system in early May, giving users a month to adjust before the June 1 rollout.

Key Points
  • GitHub Copilot shifts to token-based AI Credits pricing on June 1, 2026
  • Base prices stay at $10/month (Pro) and $39/month (Pro+), but credits match dollar value
  • Agentic coding sessions will consume credits, potentially increasing costs for heavy users

Why It Matters

This pricing shift signals that AI coding tools are moving to consumption-based models, potentially raising costs for developers.