FeynmanBench: Benchmarking Multimodal LLMs on Diagrammatic Physics Reasoning

TL;DR

FeynmanBench is a comprehensive benchmark designed to evaluate multimodal large language models' ability to perform complex diagrammatic reasoning in physics, focusing on Feynman diagrams and their associated physical laws.

Contribution

This work introduces the first benchmark centered on Feynman diagram tasks, supporting large-scale evaluation of AI's physics-grounded reasoning capabilities.

Findings

State-of-the-art MLLMs show systematic failures in physics constraints.

The benchmark includes over 2000 tasks across various fundamental interactions.

Automated pipeline ensures reproducible and diverse diagram generation.

Abstract

Breakthroughs in frontier theory often depend on the combination of concrete diagrammatic notations with rigorous logic. While multimodal large language models (MLLMs) show promise in general scientific tasks, current benchmarks often focus on local information extraction rather than the global structural logic inherent in formal scientific notations. In this work, we introduce FeynmanBench, the first benchmark centered on Feynman diagram tasks. It is designed to evaluate AI's capacity for multistep diagrammatic reasoning, which requires satisfying conservation laws and symmetry constraints, identifying graph topology, converting between diagrammatic and algebraic representations, and constructing scattering amplitudes under specific conventions and gauges. To support large-scale and reproducible evaluation, we developed an automated pipeline producing diverse Feynman diagrams along with verifiable topological annotations and amplitude results. Our database spans the electromagnetic, weak, and strong interactions of the Standard Model, encompasses over 100 distinct types and includes more than 2000 tasks. Experiments on state-of-the-art MLLMs reveal systematic failure modes, including unstable enforcement of physical constraints and violations of global topological conditions, highlighting the need for physics-grounded benchmarks for visual reasoning over scientific notation. FeynmanBench provides a logically rigorous test of whether AI can effectively engage in scientific discovery, particularly within theoretical physics.