AI 学会看题下菜碟：简单题秒答，难题才深思
大模型现在有个毛病：简单题绕圈子，难题反而想不够。这篇研究给推理模型装了个“难度感知”的开关——先看题目难不难，容易的直接给答案，难的才启动深度推理。做法是拿预计算的准确率和 token 消耗当参考，用奖励机制引导模型自己学会分配思考长度，不需要额外训练奖励模型。在数学基准上，AIME24 的通过率涨了 10%，token 用量降了 27.9%；AIME25 上 Pass@3 到 40%，比单纯压缩或路由的方案都强。它不是你明天就能用的工具，但指向一个趋势：AI 的“聪明”不再只是答得对，还包括知道什么时候该省力气。
📄 原文摘要(英文)
Large Reasoning Models (LRMs) achieve strong performance on complex tasks but exhibit systematic inefficiency: they often overthink easy problems and underthink hard ones. Existing approaches based on uniform length penalties or rigid routing incur an efficiency tax, trading reduced computation on easy instances for accuracy loss on hard instances. We formulate efficient reasoning as an instance-adaptive computation allocation problem and propose When2Think, a post-training framework for hybrid reasoning that dynamically allocates computation based on problem difficulty. Our method introduces Instance-level Difficulty-Aware Control (IDAC), a reward-shaping mechanism that leverages pre-computed reference statistics (accuracy and token usage) to regulate reasoning depth. Combined with verifier-based rewards and batch-wise standardized advantages, IDAC enables stable critic-free optimization without learned reward models or online reference-model queries. When2Think encourages direct answering on easy instances while preserving extended reasoning on hard instances, thereby learning when to use System 1 (NoThink) versus System 2 (Think). Experiments on mathematical benchmarks demonstrate improved accuracy-efficiency trade-offs: on AIME24, Pass@3 increases by 10.0% while token usage is reduced by 27.9% relative to the base model, and on AIME25, When2Think achieves 40.0% Pass@3, outperforming compression and routing-only baselines.