GitHub Copilot Code Review AI Credits and Billing Guide
On this page
Quick answer
Copilot code review has two cost layers: AI credits for the review and GitHub Actions minutes for agentic context gathering and tool use. GitHub currently estimates a typical review at $0.05–$1 in AI credits for Lite and $0.25–$5 for Balanced; those ranges can change and exclude Actions minutes.
For automatic bot-authored pull requests, identify who triggered the workflow or name a designated billing owner. Users without a Copilot license or included code-review allowance consume organization- or enterprise-paid additional usage when the required policies are enabled.
Cost model
| Cost layer | Main driver | Evidence to capture |
|---|---|---|
| AI credits | Model interaction, tokens, PR size, instructions, effort level | Credits, effort, author or requester, billing owner, review outcome |
| Actions minutes | Agentic context gathering and tool use | Runner type, duration, fallback, workflow result, hosted-runner rate |
| Human review | Validation, corrections, missed issues, disposition | Reviewer time, accepted findings, false positives, escaped issues |
| Downstream work | Tests, remediation, reruns, agent-applied suggestions | Commits, test results, security findings, approvals, rollback |
Do not reduce the business case to the AI-credit line. A cheap review that adds false positives or misses a critical issue can cost more in reviewer time and risk.
Licensed versus unlicensed usage
For an automatically requested review, GitHub normally attributes AI-credit consumption to the pull-request author. A manual request is attributed to the requesting user.
When a pull request is created by GitHub Actions or a bot, GitHub says usage applies to the identifiable user who triggered the workflow or a designated billing owner. Members without a Copilot license and users on a plan without code review have no monthly allowance for it; when enabled, their manual and automatic reviews are paid additional usage billed to the organization or enterprise.
Administrators must enable AI credits paid usage before enabling Allow members without a Copilot license to use Copilot code review in GitHub.com. Preserve the exact policy scope and note that unlicensed access is for GitHub.com, not IDE review.
Lite versus Balanced
GitHub describes Lite as the default fast, targeted review and Balanced as higher-reasoning analysis for complex logic, security-sensitive code, and cross-service changes. Balanced uses more AI credits and can consume somewhat more Actions minutes.
Set an organization default by repository risk rather than prestige. A useful pilot compares:
- accepted findings per review;
- false-positive and incorrect-resolution rate;
- issues found later by humans, tests, or incidents;
- AI credits and Actions minutes;
- elapsed time and reviewer effort;
- runner fallback and failed-review rate.
Budget controls
GitHub says code review is blocked with other AI-credit-consuming features when the applicable user budget, enterprise limit, or cost-center spending limit is exhausted. Test that denial path before relying on automatic review as a required control.
Use a budget owner, warning thresholds, reporting cadence, and a fallback reviewer. A spending cap can stop usage, but it does not prioritize the most important pull requests or provide human coverage after the block.
Runner choice
Standard GitHub-hosted runners are the default for agentic capabilities. Larger hosted runners have a higher per-minute price. Self-hosted runners do not consume GitHub Actions minutes, but they carry their own infrastructure, isolation, patching, secrets, network, observability, and capacity costs.
If hosted runners are disabled or a workflow fails, GitHub can fall back to a limited review. Record that fallback so a lower-cost result is not mistaken for the intended context-rich review.
Use the bot and large-PR guide for operational behavior and the readiness tool for a combined policy, budget, context, and validation gate.
Frequently asked questions
What does a Copilot code review cost?
GitHub currently estimates about $0.05 to $1 of AI credits for a typical Lite review and $0.25 to $5 for Balanced. Pull-request size, instructions, and model changes affect usage, and Actions minutes are separate.
Who pays for an automatic review of a bot-authored pull request?
GitHub says usage can apply to an identifiable user who triggered the workflow or to a designated billing owner. For an author without an included allowance, paid additional usage is billed to the organization or enterprise.
Are GitHub Actions minutes included in the AI-credit estimate?
No. AI credits pay for the model interaction; Actions minutes cover agentic context gathering and tool use. Larger hosted runners cost more per minute, while self-hosted runners do not consume Actions minutes.
What happens when a Copilot AI-credit budget is exhausted?
GitHub says code review and other AI-credit-consuming features are blocked when the applicable user, enterprise, or cost-center spending limit is exhausted.
Official sources
- GitHub Docs: About Copilot code review
- GitHub Docs: AI-credit billing for organizations and enterprises
- GitHub Docs: Models, pricing, and code-review usage
- GitHub Docs: Budgets for usage-based billing
Source check: August 28, 2026. Price estimates, effort levels, policy labels, allowances, and runner rates are volatile; confirm the live billing pages and your account before a rollout.