NewtonBench: Benchmarking Generalizable Scientific Law Discovery in LLM Agents
Published in ICLR 2026, 2026
Authors: Tianshi Zheng*, Kelvin Kiu-Wai Tam*, Newt Hue-Nam K. Nguyen*, Baixuan Xu, Zhaowei Wang, Jiayang Cheng, Hong Ting Tsang, Weiqi Wang, Jiaxin Bai, Tianqing Fang, Yangqiu Song, Ginny Y. Wong, Simon See
*Equal contribution (co-first authors)
NewtonBench addresses limitations in prior benchmarks for LLM-driven scientific law discovery by combining counterfactual law shifts (systematic alterations of canonical physics laws) with interactive model discovery in simulated experimental environments.
Recommended citation: Tianshi Zheng*, Kelvin Kiu-Wai Tam*, Newt Hue-Nam K. Nguyen*, et al. (2026). "NewtonBench: Benchmarking Generalizable Scientific Law Discovery in LLM Agents." ICLR. *Equal contribution.
Download Paper