GitHub Copilot adds auto model selection tiers, code review enhancements, and Sentry integration

GitHub Copilot introduced three auto model selection tiers—efficiency, balance, and intelligence—that let users prioritize cost, quality, or response time while drawing from the same model pool; this feature is rolling out in VS Code, the Copilot CLI and the Copilot app. Code review improvements include automatic resolution of addressed comments in subsequent reviews, suggestions for commit messages when applying code suggestions, support for shell tools to validate changes, and Lite reviews that combine findings from multiple agents. The Copilot app now integrates Sentry, providing a canvas that shows production crash reports, related context and enables developers to investigate, validate fixes and prepare pull requests. Business and Enterprise users gain generally available usage metrics—daily active users, session counts and message totals—in the VS Code Agents window, while a public preview lets Copilot suggest allowed values for repository custom properties with policy‑controlled enablement. A budget‑request feature for higher AI credit limits is generally available for Copilot Business and Enterprise plans with usage‑based billing, though it is not offered to enterprises with managed users. Additional agent features include running agents in local Dev Containers (requiring Docker and a supported configuration), automatic cleanup of inactive sessions in preview, and the ability to create pull requests directly from the Agents window. The updates were announced on September 18 2026 covering the September 14 release; availability varies, with some features rolling out gradually, others generally available, and some in public preview. All details are based on GitHub’s own changelog announcement.