GitHub Copilot in VS Code: Agents Can Run Model Comparisons with Compare Models

GitHub CopilotView original changelog

In VS Code 1.146, enabling Compare Models and its agent tools lets agents find available models, launch comparisons, and summarize results. The feature is experimental.

Key Takeaways

  • Agents can launch model comparisons themselves.
  • They can find available models.
  • They read results and summarize findings.
  • Compare Models must be enabled.
  • Agent tools must be turned on in settings.
  • The feature is experimental in VS Code 1.146.

Agent-driven model comparisons

With Compare Models enabled and the agent tools turned on in settings, agents in VS Code can find available models, launch comparisons, and read the results to summarize findings. This is experimental and part of the VS Code 1.146 release.

VS Code Copilot Agents Can Run Model Comparisons | Yet Another Changelog