Large Reasoning Models (LRMs) have revolutionized complex problem-solving, yet they exhibit a pervasive "overthinking", generating unnecessarily long reasoning chains. While current solutions improve token efficiency, they often sacrifice fine-grained control or risk disrupting the logical integrity of the reasoning process. To address this, we introduce Stepwise Adaptive Thinking (SAT), a framework that performs step-level, difficulty-aware pruning while preserving the core reasoning structure. SAT formulates reasoning as a Finite-State Machine (FSM) with distinct thinking modes (Slow, Normal, Fast, Skip). It navigates these states dynamically using a lightweight Process Reward Model (PRM), compressing easy steps while preserving depth for hard ones. Experiments across 9 LRMs and 7 benchmarks show that SAT achieves up to 40% reduction in reasoning tokens while generally maintaining or improving accuracy.
@article{arxiv.2604.07922,
title = {SAT: Balancing Reasoning Accuracy and Efficiency with Stepwise Adaptive Thinking},
author = {Weiyang Huang and Xuefeng Bai and Kehai Chen and Xinyang Chen and Yibin Chen and Weili Guan and Min Zhang},
journal= {arXiv preprint arXiv:2604.07922},
year = {2026}
}