大模型不用每次重新想，答案能缓存
大模型每次解题都从零开始，哪怕同一道题换个数字。这篇论文发现：一旦某个任务被推理过，把它的解法套到新例子上，几乎不需要大模型参与。他们从推理轨迹里提取出可执行的「推理缓存」，在数学和科学难题上，效果和顶级模型持平，但调用成本便宜 16 到 291 倍，而且完全可解释、能持续更新、不怕输入扰动。它不是你明天能用上的，但它指向一个方向：AI 的聪明可以沉淀成可复用的资产，而不是每次重新烧钱。
📄 原文摘要(英文)
Large language models reinvent the wheel all the time. They conflate task-solving with the execution of the solutions, which forces them to solve every problem from scratch. We show that once a task has been reasoned through, the application of its solution to a new problem instance can be done very efficiently and with little involvement from language models. We introduce agentic reasoning caching, a method for constructing robust CPU-executable agentic reasoning caches (ARCs) from LLM reasoning traces. ARCs show on-par performance with frontier LLMs across a range of difficult mathematical and scientific tasks, are fully explainable, continually trainable, robust to input perturbations, and 16-291x cheaper to invoke than the source model.