Starting June 1, 2026, GitHub will switch GitHub Copilot’s AI billing from a flat-rate, unlimited request model to a usage-based system metered by AI tokens, introducing GitHub AI Credits to better align costs with actual resource consumption. While subscription prices remain unchanged, users may face variable costs based on usage, with some AI features remaining unlimited, reflecting GitHub’s effort to manage escalating AI infrastructure expenses sustainably amid growing demand.
Microsoft’s GitHub is transitioning its AI billing model for GitHub Copilot from a request-based system to a usage-based, metered billing approach starting June 1, 2026. This change comes as GitHub acknowledges that the previous “all-you-can-eat” style subscription model, which allowed unlimited premium requests regardless of complexity, has become financially unsustainable. The company has been absorbing escalating inference costs, especially since complex AI requests can be significantly more resource-intensive than simpler ones, leading to losses under the flat-rate request billing system.
Under the new usage-based billing model, GitHub will meter AI usage based on tokens, which are units representing small chunks of text processed by the AI. Different AI models consume tokens at varying rates, so GitHub has introduced a virtual currency called GitHub AI Credits, valued at $0.01 each, to standardize billing. Subscribers will receive a monthly allotment of these AI Credits depending on their plan, with options to purchase additional credits if needed. This system aims to more accurately align costs with actual AI resource consumption, reflecting input, output, and cached tokens.
GitHub plans to provide users with a preview billing experience in early May 2026 to help them estimate their upcoming costs before the full transition. Despite the shift, some AI features like code completions and Next Edit Suggestions will remain unlimited even after users exhaust their AI Credit allotment. Subscription prices themselves will remain unchanged, but the new billing model introduces variability in costs depending on usage, which may be difficult for users to predict precisely due to the non-deterministic nature of AI token consumption.
This move by GitHub follows a broader industry trend where AI service providers are adjusting pricing models in response to soaring demand and infrastructure costs. Other companies like Anthropic, Google, and OpenAI have also implemented usage limits or introduced higher-priced subscription tiers to manage capacity and costs. Cloud providers such as AWS and Azure have faced similar challenges, highlighting the growing pains of scaling AI services sustainably amid rapid adoption and experimentation with AI agents.
For existing annual subscribers, GitHub offers options to cancel with a pro-rated refund or downgrade to a free plan upon subscription expiration, as these annual plans will not be renewed. Additionally, premium AI models will see significant price multiplier increases under the new system, reflecting their higher computational costs. Overall, GitHub’s shift to metered AI billing represents a strategic effort to balance user demand with financial viability, moving away from the unlimited usage model that proved costly to maintain.