GitHub Copilot June 20 update — smarter prompt caching, deferred tool loading, Auto model selection routes tasks to the right model in VS Code and beyond
GitHub Copilot's June 20 update improves efficiency through smarter prompt caching, deferred tool loading, and Auto model selection that routes tasks to the appropriate model in VS Code and other surfaces. The Auto model selection is the substantive feature — automating the per-task model-routing decision that developers previously made manually based on task complexity and cost considerations.
The substantive piece is the Auto model selection feature operationalizing what enterprise procurement has been advocating. Pre-Auto-model-selection Copilot users manually selected which model to use per task — Claude Fable 5 for complex multi-file refactors, GPT-5.5 for general coding, Opus 4.8 for reasoning-heavy debugging. The Auto selection makes the per-task routing decision automatically based on task characteristics and configured cost preferences.
The competitive read against the June 17 Copilot Desktop App GA is that GitHub is now shipping efficiency-and-routing improvements rapidly alongside the agent-control-plane positioning. The H2 2026 Copilot competitive position improves on multiple axes simultaneously — agent control plane (Desktop App), per-task model routing (Auto selection), prompt-caching efficiency (June 20 update), Fable 5 integration (June 9).
Releasebot — GitHub Release Notes - June 2026 Latest Updates → · GitHub Blog — GitHub Copilot code review and Auto model selection updates →