---
language:
- en
license: apache-2.0
size_categories:
- n=520
task_categories:
- question-answering
- text-generation
pretty_name: CMPhysBench
tags:
- Condensed Matter Physics
- physics
- benchmark
---

# CMPhysBench: A Benchmark for Evaluating Large Language Models in Condensed Matter Physics

> 🎉🎉🎉 This paper is accpeted by ICLR 2026.

[![Paper](https://img.shields.io/badge/Paper-B31B1B?logo=arxiv)](https://arxiv.org/abs/2508.18124)   [![Code](https://img.shields.io/badge/Code-8A2BE2?logo=github)](https://github.com/CMPhysBench/CMPhysBench)   [![Data](https://img.shields.io/badge/Data-FFD700?logo=huggingface)](https://huggingface.co/datasets/weidawang/CMPhysBench)   [![License](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](https://github.com/CMPhysBench/CMPhysBench/blob/main/LICENSE)

We introduce **CMPhysBench**, designed to assess the proficiency of Large Language Models (LLMs) in **C**ondensed **M**atter **Phys**ics, as a novel **Bench**mark. CMPhysBench is composed of more than 520 graduate-level meticulously curated questions covering both representative subfields and foundational theoretical frameworks of condensed matter physics, such as magnetism, superconductivity, strongly correlated systems, etc. To ensure a deep understanding of the problem-solving process,we focus exclusively on calculation problems, requiring LLMs to independently generate comprehensive solutions. Meanwhile, leveraging tree-based representations of expressions, we introduce the Scalable Expression Edit Distance (SEED) score, which provides fine-grained (non-binary) partial credit and yields a more accurate assessment of similarity between prediction and ground-truth. Our results show that even the best models, Grok-4, reach only 36 average SEED score and 28% accuracy on CMPhysBench, underscoring a significant capability gap, especially for this practical and frontier domain relative to traditional physics. 

<div align="center">
  <img src="https://raw.githubusercontent.com/CMPhysBench/CMPhysBench/main/imgs/CMPhysBench.png" width="1000"/>
</div>


## Citations

```bibtex
@article{wang2025cmphysbench,
  title={CMPhysBench: A Benchmark for Evaluating Large Language Models in Condensed Matter Physics},
  author={Wang, Weida and Huang, Dongchen and Li, Jiatong and Yang, Tengchao and Zheng, Ziyang and Zhang, Di and Han, Dong and Chen, Benteng and Luo, Binzhao and Liu, Zhiyu and others},
  journal={arXiv preprint arXiv:2508.18124},
  year={2025}
}
```