PhysUniBench: A Multi-Modal Physics Reasoning Benchmark at Undergraduate Level

TL;DR

PhysUniBench is a comprehensive multimodal benchmark designed to evaluate and enhance AI models' physics reasoning at the undergraduate level, covering diverse sub-disciplines and problem types with visual diagrams.

Contribution

This paper introduces PhysUniBench, a large-scale, rigorously curated physics reasoning benchmark for multimodal large language models, filling a gap in standardized undergraduate physics evaluation.

Findings

Current models struggle with physics reasoning, achieving only 51.6% accuracy.

Models find multi-step problems and diagram interpretation particularly challenging.

PhysUniBench provides a new standard for assessing AI's physics understanding.

Abstract

Physics problem-solving is a challenging domain for AI models, requiring integration of conceptual understanding, mathematical reasoning, and interpretation of physical diagrams. Existing evaluations fail to capture the full breadth and complexity of undergraduate physics, whereas this level provides a rigorous yet standardized testbed for pedagogical assessment of multi-step physical reasoning. To this end, we present PhysUniBench, a large-scale multimodal benchmark designed to evaluate and improve the reasoning capabilities of multimodal large language models (MLLMs) specifically on undergraduate-level physics problems. PhysUniBench consists of 3,304 physics questions spanning 8 major sub-disciplines of physics, each accompanied by one visual diagram. The benchmark includes both open-ended and multiple-choice questions, systematically curated and difficulty-rated through an iterative process. The benchmark's construction involved a rigorous multi-stage process, including multiple roll-outs, expert-level evaluation, automated filtering of easily solved problems, and a nuanced difficulty grading system with five levels. Through extensive experiments, we observe that current models encounter substantial challenges in physics reasoning, where GPT-5 achieves only 51.6% accuracy in the PhysUniBench. These results highlight that current MLLMs struggle with advanced physics reasoning, especially on multi-step problems and those requiring precise diagram interpretation. By providing a broad and rigorous assessment tool, PhysUniBench aims to drive progress in AI for Science, encouraging the development of models with stronger physical reasoning, problem-solving skills, and multimodal understanding.