Armin theorizes that this is because more recent Anthropic models have been specifically trained (presumably via Reinforcement Learning) to better use the edit tools that are baked into Claude Code. This has the unfortunate effect that other coding harnesses, such as Pi, may find that their own custom edit tools are more likely to be used incorrectly.

Claude’s edit tool uses search and replace. OpenAI’s Codex uses an apply_patch mechanism instead, and OpenAI have talked in the past about how their models are trained to use that tool effectively.

Does this mean third-party coding harnesses like Pi should implement multiple edit tools just so they can use the one with the best performance for the underlying model the user has selected?

Source: Better Models: Worse Tools

This is an awful trend if this sticks. Models were supposed to be generic - however, if this trend continues, third party harnesses will once again face hurdles with the frontier models.

One can hope that frontier labs are smart enough to know that this is a terrible long term trend as a generic coding model has far better payoffs in the long term than a first party harness lock-in.