GitHub Makes Lite and Balanced Copilot Code Reviews Generally Available
Developers can choose review depth per pull request, while organizations set inherited defaults and account for different resource use.
GitHub made Lite and Balanced effort levels for Copilot code review generally available on August 7. The two settings replace the Low and Medium names used during public preview, and existing configurations carry over automatically.
The change gives developers and administrators a visible control over how much analysis Copilot applies to a pull request. It matters most to teams trying to reserve deeper automated review for complex or sensitive work while keeping routine reviews faster and less resource-intensive.
Lite becomes the default review path
GitHub describes Lite as its standard review level for targeted feedback on issues such as bugs, security weaknesses and style inconsistencies. It is the default option and is intended for routine changes where speed matters more than exhaustive analysis.
Balanced routes a pull request to what GitHub calls a higher-reasoning model for longer analysis of complex logic, security-sensitive code and changes that cross services. The documentation says Balanced consumes more AI credits than Lite and may also use slightly more GitHub Actions time. That makes the setting an operational and budget choice, not merely a label in the review interface.
Teams can choose at review, repository and organization level
A developer can select Lite or Balanced when requesting one Copilot review without changing the standing default. Organization administrators can set a default that repositories inherit, while repository administrators can override that organization choice for a specific repository.
After a review runs, GitHub displays the effort level in the pull-request overview and timeline. That visibility gives teams a basic audit trail for which depth setting produced a particular set of comments.
The changelog lists Copilot Pro, Pro+, Max, Business and Enterprise as supported plans. GitHub's broader documentation also notes that organization policy must enable code review for organization-managed access.
Runner policy changes the available context
Copilot code review can use GitHub Actions for agentic features including broader project-context gathering and handing suggestions to Copilot's cloud agent. GitHub says a review can still be generated if those Actions workflows are unavailable, but it will not include those extra capabilities.
Organizations that disable GitHub-hosted runners therefore need to distinguish between getting a review and getting the fuller agentic review path. GitHub documents self-hosted runners as an alternative for that context-gathering work.
What general availability does not establish
The official material explains configuration, availability and resource behaviour, but it does not provide an independent comparison of defect detection, false positives or security findings between Lite and Balanced. A higher-reasoning route should not be read as proof that a review is complete or that human review and testing can be removed.
For engineering leaders, the useful next checkpoint is local evidence: compare finding quality, review latency, developer rework and resource use across similar pull requests. That would show whether the extra analysis is improving outcomes in a specific codebase rather than only increasing review effort.
Status
Confirmed. Internal confidence: medium. GitHub's changelog and documentation establish the product change, while independent quality testing was not available in this run.
Sources
Update note: Last reviewed 2026-08-14. We will revise this post if GitHub changes plan access, effort behaviour or the documented usage model.
Sources
- GitHub changelog announcement — official
- GitHub Copilot code review documentation — official
Drafted with AI assistance from source briefs; reviewed for citation completeness and label accuracy.