GitHub Copilot Code Review API Arrives as Balanced Becomes Default

What changed

GitHub has opened up Copilot code review to programmatic control. Developers can now request a Copilot review through the REST and GraphQL APIs, and each request can optionally specify its own review effort level 4. The stated aim is to let reviews start from the scripts, workflows and internal tools teams already run 41. The feature is generally available on Copilot Pro, Pro+, Max, Business and Enterprise plans 41.

Alongside the API, GitHub confirmed a second change that is arguably more consequential for budgets. The "Default" review effort setting now resolves to Balanced rather than Lite, for both new and existing repositories and organizations 4. GitHub announced this on August 28, 2026, and it took effect September 28 4. Teams that had explicitly picked Lite kept that setting. Everyone who left the option on Default moved to the higher tier automatically 4.

Why the pairing matters

Each change is reasonable on its own. Together, they change the economics of AI review.

The API closes a gap that frustrated automation-minded teams. Before native support, many could automate the first review request but struggled to reliably re-request a bot review after new commits. That often meant someone had to step in through the web interface 2. With the API, it becomes straightforward to wire reviews into common triggers. Examples include opening a pull request, adding a label like "needs-ai-review," pushing new commits, or escalating from Lite to Balanced only for high-risk changes 2.

Easier triggering means more reviews, and the reviews now cost more by default. Since June 1, 2026, each Copilot review consumes AI credits priced at $0.01 apiece, plus GitHub Actions minutes on private repositories 3. GitHub's own estimates put a Lite review at roughly $0.05 to $1 in credits. A Balanced review runs about $0.25 to $5 3. The upper bound at Balanced is five times Lite's, and the lower bound is likewise five times higher. A pipeline that re-reviews on every push, inheriting the new default, could burn through credits far faster than a team expects. One analysis frames the effort default, AI credits and Actions minutes together as a "hidden cost multiplier" 2.

What GitHub hasn't said

GitHub's changelog leaves some practical questions open. It does not say whether an API request interacts with automatic review settings. It also does not say what permissions or prerequisites a request requires 1. GitHub's documentation already described ways to request a review in code. One example is adding Copilot as a reviewer via the REST API or with a command such as gh pr edit 123 --add-reviewer @copilot 13. So the new piece is mainly per-request effort control and formal support, not the bare ability to call a review from a script.

It's also worth remembering what Copilot review is and isn't. It posts line comments with severity labels and suggested fixes that can be committed directly. However, it cannot block a merge 3. Teams building gating logic around it will need their own checks on top.

Reading the move

The outlets broadly agree on the facts and differ mainly in emphasis. GitHub presents the update as a convenience and quality improvement, with Lite available to anyone who opts out 4. Independent coverage focuses on cost, noting that the API and the pricier default arrived within days of each other 23.

In my reading, the cost concern is the right lens for most organizations. Defaulting to Balanced probably does improve the review quality many users see. Making reviews programmable is a genuine workflow win. But the defaults now favor heavier spending at exactly the moment automation makes volume easy to scale. Teams that set Default once and forgot about it are already paying Balanced rates.

The practical response is not to avoid the API but to be deliberate. Teams can take four concrete steps:

• Check which effort level each repository or organization actually resolves to.
• Set Lite explicitly where review depth matters less.
• Use the per-request effort parameter to reserve Balanced for risky changes.
• Monitor credit and Actions consumption after wiring reviews into CI.

The escalation pattern described in the coverage, starting light and stepping up only when warranted, looks like the sensible template 2. Teams that skip this housekeeping may find the bill for programmable review arriving before its benefits are clear.