Comment by procinct

The models are the same. The agent implementation is different. I can confirm Claude Code performs much better than GH Copilot with the same Claude models.