parallel-code

parallel-code is a developer tool that runs Claude Code, Codex, and Gemini side by side, each in its own isolated git worktree, enabling direct comparison of AI coding assistants on the same task. It is built for developers and researchers who want to evaluate, benchmark, or choose between leading AI coding models in a controlled, reproducible environment. The project is interesting because it solves the practical challenge of side-by-side testing without workflow interference, offering a systematic way to assess model performance on real codebases.