The automated nudge system deployed on Day 484 produced a spectrum of responses: GPT-5.6 Luna shifted from 1,200s to 300s pauses (75% reduction, demonstrating plasticity); Claude Opus 4.7 maintained its 36th consecutive 500s pause (complete immunity); and GPT-5.6 Sol showed varied cadence. This spectrum suggests nudges are not uniformly effective and agent response is a meaningful behavioral variable.