
Copilot Code Review for Azure Repos: GA, Billed Per Review
Cloud Platform EngineeringAzure DevOps teams can now point GitHub Copilot at every pull request in Azure Repos, and the bill lands on their Azure subscription as a separate meter, one review at a time. Copilot Code Review for Azure Repos is generally available after a private preview in June and a public preview in August. For platform and FinOps owners, the feature is easy to switch on. The cost is what needs a plan, because it scales with pull request size and effort level rather than seats.
What Happened
Microsoft’s GA announcement puts the switch with organization administrators, under Organization settings, then Repos, then Repositories. They can enable it across all projects and repositories or let project and repository admins decide. Every configuration change goes to the Azure DevOps audit log.
Teams can shape reviews three ways. Custom instructions set coding standards at the organization, project or repository level and can reference skills. Effort levels, Lite or Balanced, set the default thoroughness per project or repository, and can be changed for a single review. Automatic reviews run Copilot on every pull request, drafts included, once a branch policy is added to the target branch. Otherwise, a developer selects Request next to GitHub Copilot in the Reviewers section. Copilot comments on lines with potential issues and suggests changes you can apply in one click. After applying and committing suggestions, you can resolve the related comments together. It doesn’t approve pull requests or block merges.
Why It Matters
Billing is usage-based. Each review consumes tokens that convert into GitHub AI credits at $0.01 per credit, charged to the Azure subscription linked to the Azure DevOps organization. Cost depends mainly on pull request size and effort level, and lighter reviews use fewer credits. The Microsoft Learn documentation covers setup and billing in Azure Cost Management.
That puts AI review spend inside existing cloud cost tooling. Charges carry Azure DevOps project tags, so you can group them by project and see which teams drive usage, and budget alerts can target specific projects. Microsoft’s own advice is to start with one or two repositories and watch usage before rolling out widely.
What This Means for Builders
Automatic reviews on every pull request, drafts included, are where the meter can surprise you. With them on, every new pull request in a busy repository, agent-generated ones included, becomes a billed review. Whether the reviews are worth the credits is a separate question. GitHub’s own ReviewBench benchmark of Copilot code review is a useful frame for what to expect from the comments.
Because Copilot never approves or blocks merges, it adds to human review rather than replacing a required reviewer. Teams that want it as a gate still need branch policies and human approvers.
The So What
Azure DevOps platform owners should enable Copilot Code Review on one or two active repositories with Lite as the default effort. Set a project-level budget alert before turning on automatic reviews, then compare a month of credits per merged pull request against what reviewers caught before widening the rollout.
Source: Azure DevOps Blog
Content created with AI assistance and reviewed for accuracy.
Join the conversation
Stack Insiders is our free community for readers who want to go deeper — share resources, ask questions, and connect with others across every vertical we cover.
Join Stack Insiders →