Frontier models agree on financial decisions 95% of the time but follow the same tool path only 77%.

Raffi Khatchadourian's replay benchmark measures behavioral instability through three channels that need no access to hidden reasoning text: tool-call trajectories, evidence contacts, decision concentration (arXiv 2607.20491). Across 8,127 replay episodes over 10 models and 3 financial tasks, that 18-point gap (95% CI [0.14, 0.22]) is completely invisible to outcome-only evaluation, and over 55% of high-agreement case groups show meaningful trajectory divergence. It separates agents into pattern matchers that collapse to one output regardless of input, stable executors, and trajectory divergers. Log the path, not just the answer.