Quantifying and Understanding Uncertainty in Large Reasoning Models

ArXi:2604.13395v1 Announce Type: new Large Reasoning Models (LRMs) have recently nstrated significant improvements in complex reasoning. While quantifying generation uncertainty in LRMs is crucial, traditional methods are often insufficient because they do not provide finite-sample guarantees for reasoning-answer generation. Conformal prediction (CP) stands out as a distribution-free and model-agnostic methodology that constructs statistically rigorous uncertainty sets. However, existing CP methods ignore the logical connection between the reasoning trace and the final answer.