# CPU-ARC: Agentic Reasoning Caching for on-CPU Reasoning Reuse

**Authors:** Peter Belcak, Greg Heinrich, Saurav Muralidharan, Shizhe Diao, Pavlo Molchanov  
**Lab:** NVIDIA · **Published:** 5 Nov 2026 · **Upvotes:** 10  
**Links:** [arXiv](https://arxiv.org/abs/2611.09891) · [PDF](https://arxiv.org/pdf/2611.09891) · [This page](https://omen-bit.github.io/ai-paper-daily/paper/2611.09891.html)

## Abstract

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.

---

Notes written by AI from the paper, numbers copied from the abstract. Check the paper before you rely on them. From AI Papers Daily.