GitHub Copilot Code Review API Adds Effort Control as Balanced Becomes Default

What changed​

1. Review requests through the API. GitHub says you can request a Copilot review through the supported REST and GraphQL APIs. You can optionally set the effort level for that one review. GitHub's stated goal is to let reviews start from scripts, workflows and internal tools.

The changelog doesn't say whether the API turns automatic review on. It also doesn't say what permissions or prerequisites a request needs.

GitHub's docs describe an existing way to request a review in code. You can request a review through the GitHub REST API by requesting copilot-pull-request-reviewer[bot] as a reviewer. Before this announcement, a community issue in a third-party repository said the REST requested_reviewers endpoint, GraphQL requestReviews/requestReviewsByLogin, the GitHub MCP request_copilot_review tool, and gh pr edit --add-reviewer had no effort parameter. That is one user's account, not GitHub documentation. It does suggest that the new per-request effort option fills a gap automation users had noticed. Check GitHub's current API reference for the exact parameter names before you script against it.

2. Balanced is the default. The changelog says the Default setting now resolves to Balanced for repositories and organizations using Copilot code review. The change took effect on September 28, 2026. GitHub had announced it on August 28. If you explicitly chose Lite, that choice was kept.

Why the default matters​

Balanced and Lite differ in depth and cost. GitHub's documentation describes them this way:

• Lite gives fast, targeted feedback on common issues such as bugs, security vulnerabilities and style inconsistencies.
• Balanced sends the pull request to a higher-reasoning model for longer analysis. It is aimed at complex logic, security-sensitive code and cross-service changes. It uses more AI credits and may use marginally more GitHub Actions minutes.

GitHub's documentation estimates typical AI-credit use at $0.05 to $1 per Lite review and $0.25 to $5 per Balanced review. Usage generally rises with pull request size and repository custom instructions. The estimates can change as models evolve, and they exclude Actions minutes.

Anything that relied on Default, including automated review requests, now gets the deeper and costlier level unless a more specific setting says otherwise. I can't say how much that raises any particular team's bill. That depends on pull request volume, size and custom instructions.

How Copilot picks the effort level​

GitHub's documentation lists the order in which Copilot checks for an effort level. It uses the first one that applies:

1. The level chosen when the review is requested.
2. A level previously used on that pull request.
3. The requestor's own setting.
4. The repository setting.
5. The organization setting, or the repository owner's setting for a user-owned repository.
6. GitHub's built-in default, which is Balanced.

An effort level supplied in an API request therefore takes precedence over every stored setting. The pull request overview comment shows the effort level used for each review run.

How to keep Lite​

Choose Lite explicitly at the level you manage, instead of leaving the setting at Default. GitHub's changelog gives these paths:

• Enterprise: enterprise settings, then AI controls, then Agents, then Copilot code review.
• Organization: organization settings, then Copilot, then Code review.
• Repository: repository settings, then Copilot, then Code review.
• Personal: profile picture, then Copilot settings, then Copilot, then Code review.

GitHub says each level can override the one above it. GitHub's configuration docs say you can return to the built-in default by selecting the Default option at the top of the menu. A September 23 GitHub changelog added an enterprise-wide default review effort of Lite, Balanced or the GitHub default. It also says organizations and repositories can still set their own overrides.

Your personal effort setting is independent of automatic review. GitHub's docs say turning automatic review off doesn't clear your effort setting or stop it applying to reviews you request manually.

Practical takeaways​

• Audit your settings. Find repositories and organizations still on Default. They are now on Balanced.
• Set effort per request. GitHub's docs suggest Lite for routine changes where fast feedback matters most. They suggest Balanced for security-sensitive code, multi-service pull requests and repositories with strict quality standards.
• Test automation first. One third-party guide advises trying the reviewer request on a sandbox repository before adding it to a workflow. That is sensible advice for any new API integration.
• Keep humans in the loop. GitHub's docs say Copilot isn't guaranteed to catch every problem and its feedback should be validated. They also say to supplement it with human review.
• Watch budgets. For Business and Enterprise, GitHub's docs say that when a user or enterprise budget is exhausted, code reviews are blocked along with other AI-credit features.

Bottom line​

The API support is the bigger change for teams that automate their pull request process. The default change is the one more likely to surprise a bill. Both are live now, so check your effort settings and your automation.