User Maps LLM Behaviors To Control System Tuning: Claude Overdamped, ChatGPT Overshoots

A half formed thought: has anybody talked about different LLMs in terms of control system tuning? Like to me Claude feels like it's overdamped with higher steady state error while ChatGPT feels like it randomly has higher overshoot.