Copilot code review gets API requests and a Balanced default effort level

GitHub announced on October 2, 2026, that Copilot code review can be requested through its REST and GraphQL APIs, with an optional review effort level on each request. The same changelog entry confirms that Balanced is now the default review effort level, a change that took effect on September 28, 2026. Both changes are generally available on the Copilot Pro, Pro+, Max, Business and Enterprise plans.

What changed in Copilot code review

There are two separate changes. The first is programmatic access with effort control: scripts, CI jobs and internal tools can start a review through the API and choose its effort level on each request.

The second is the default. GitHub first announced on August 28, 2026, that the Default review effort level would switch to Balanced for new and existing repositories and organizations using Copilot code review. According to the GitHub changelog entry (opens in a new tab), anyone who had explicitly picked Lite kept that setting.

Lite, Balanced and Max effort levels

GitHub's overview of Copilot code review (opens in a new tab) defines the levels:

• Lite: a standard review with fast, targeted feedback on common issues such as bugs, security vulnerabilities and style problems.
• Balanced: routes the pull request to a higher-reasoning model for longer analysis of complex logic, security-sensitive code and cross-service changes.
• Max: listed in settings as "Coming soon" and not available yet.

GitHub suggests Balanced for security-sensitive code, multi-service pull requests and repositories with strict quality standards, and Lite for routine changes where speed matters more.

What a Copilot review costs at each level

The new default changes spending. The same overview page estimates that a review typically uses $0.05 USD to $1 USD worth of AI credits on Lite, and $0.25 USD to $5 USD on Balanced. Those ranges exclude GitHub Actions minutes, which code review also uses for context gathering and tool calls. GitHub says Balanced may use marginally more Actions minutes than Lite. Larger pull requests and repository custom instructions push consumption up.

Billing follows the person who asked. For automatic reviews, AI credits are attributed to the pull request author. When another user requests a review manually, it is charged to that user. For Business and Enterprise, a user who hits their budget, or an exhausted enterprise or cost center limit, blocks code review along with other AI credit features.

How to request a review through the API

The changelog does not name the endpoints or the field for effort. GitHub's how-to for using code review says you can request a Copilot review through the REST API by adding copilot-pull-request-reviewer[bot] as a reviewer, using the request reviewers endpoint (opens in a new tab). Put together, the call looks like this:

curl -L \
  -X POST \
  -H "Accept: application/vnd.github+json" \
  -H "Authorization: Bearer <YOUR-TOKEN>" \
  -H "X-GitHub-Api-Version: 2026-03-10" \
  https://api.github.com/repos/OWNER/REPO/pulls/PULL_NUMBER/requested_reviewers \
  -d '{"reviewers":["copilot-pull-request-reviewer[bot]"]}'

A fine-grained token needs write access to the "Pull requests" repository permission. As of October 5, 2026, the REST reference for this endpoint lists only reviewers and team_reviewers as body parameters, so the name of the effort field is not documented there yet. The matching GraphQL mutation is not documented on the pages the changelog links to.

To pick an effort level, Copilot works down a fixed list and uses the first match. First comes the effort chosen at request time, then the effort already used on that pull request, then the requester's own setting. After that it checks the repository setting and the organization setting (or the owner's setting on a user-owned repository). The last fallback is GitHub's built-in default of Balanced.

How to keep Lite as your default

To stay on Lite, change Default to Lite at the level you manage. The code review configuration guide (opens in a new tab) and the changelog give these paths:

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

Each level can override the one above it. A September 23, 2026, changelog entry added the personal code review page to every Copilot plan and introduced the enterprise-wide default. Your personal effort setting also stays in force if you turn automatic review off, and it applies to reviews you request by hand.

Open questions

• The API parameter for effort and the GraphQL mutation are not documented on the pages GitHub links from the changelog.
• How AI credits are attributed when a bot or service account calls the API is not stated.
• No date is given for the Max level.
• The cost figures are GitHub's estimates, and GitHub says the ranges may change as models change.