GitHub Copilot's usage-based billing is live — what actually changed on day one

GitHub Copilot's usage-based billing went live on June 1, 2026. If you are on a monthly Pro or Pro+ plan, your account migrated automatically — no action was required. If you want the full mechanics of what changed and why, see our May 20 explainer, which covers the pricing tiers, AI Credits system, and what stays free. This article covers only what is new and observable today: what the day-one billing picture looks like, the promotional credits for Business and Enterprise accounts, and the community response.

The Business and Enterprise promo most people missed. The headline billing change is well-documented. Less covered: Copilot Business and Enterprise subscribers received a temporary promotional AI Credit allotment for June through August 2026 that is higher than the standard plan-dollar parity model. Business plans receive $30 in monthly AI Credits during the promo period, against a $19/user/month plan price. Enterprise plans receive $70 in monthly AI Credits, against a $39/user/month price. These rates exceed the dollar-for-dollar parity that monthly individual plans use. Organisations that reviewed their expected costs before June 1 now have more headroom than the standard model implied — at least through August, when the promo expires and standard parity takes effect.

The Pro+ heavy-user reality. For individual Pro+ subscribers ($39/month), the first-day billing picture is less comfortable for heavy users. Developer community analysis and third-party billing trackers indicate a Pro+ user running agent mode and frontier models aggressively can consume approximately $10–14 in AI Credits per day of typical heavy use — burning through the $39 monthly allotment in three to four days, with published overage rates applying beyond that. Light users — those whose Copilot use is primarily tab-completions and occasional Chat — are largely unaffected: code completions remain free and do not consume credits. The practical divide is agent-mode-heavy versus completion-primary workflows.

Community response: the cost-to-value debate. Developer response to the June 1 go-live has been split along usage lines, as expected. The pattern: light users who measured against their actual usage found themselves within or under their credit allotment; heavy users found the new model materially more expensive than the premium-request system it replaced. The core friction is that the premium-request system quietly subsidised token-heavy work at flat-fee pricing — the AI Credits model makes that true cost visible. That visibility is either more honest pricing or a real bill increase depending on your workflow. Cursor's usage-based Bugbot model, which went to per-usage billing around the same period, has surfaced as a comparison point for developers evaluating whether to switch tools now that pricing-model parity exists across the category.

Annual subscribers: nothing changes yet. If you are on an annual Copilot Pro or Pro+ plan, your billing model is unchanged until your plan renewal date. You remain on premium-request pricing. GitHub confirmed annual subscribers receive increased model multipliers as a transition bridge. The relevant action is to use the remaining runway before renewal to measure token consumption against the new model rates and decide whether to renew, switch plans, or evaluate alternatives. That decision is now finite — the migration is running, not approaching.

Where this leaves the AI coding tool field. The June 1 go-live completes a pricing-model shift that was already underway across the category. Claude Code bills against the Claude Pro subscription quota; Cursor bills Bugbot usage against per-review rates; GitHub Copilot now joins them. Usage-based billing is the category default — the comparison between tools is now model quality, IDE fit, and agent behaviour, not billing model. See our Cursor vs GitHub Copilot comparison and the best AI coding tools roundup for the full category picture now that pricing parity exists.