AI Paper Atlas

Research map / Large language models / Agents, code and math

HiPhO: How Far Are (M)LLMs from Humans in the Latest High School Physics Olympiad Benchmark?

Fangchen Yu, Haiyuan Wan, Qianjia Cheng, Yuchen Zhang, Jiacheng Chen, Fujun Han, Yulun Wu, Junchi Yao, Ruilizhen Hu, Ning Ding, Yu Cheng, Tao Chen and 5 more

ICML 2026 · Large language models › Agents, code and math · approach: Mathematical and structural reasoning benchmarks · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
move from text-only to text plus data figure problemsGemini-2.5-Pro · HiPhO modality typesmean normalized score (%): 86 -> 67Hurt
replace GPT-5's variable-figure score with an open-source MLLMQwen2.5-VL-72B-Instruct vs GPT-5 · HiPhO Text+Variable Figuremean normalized score (%): 75 -> 29Hurt
switch from other fields to optics problemsGemini-2.5-Pro · HiPhO physics fieldsmean normalized score (%): 89 -> 55Hurt
add extended reasoning (thinking mode) to Claude-4-SonnetClaude-4-Sonnet · HiPhO (13 Olympiads)gold medals: 2 -> 8Improved
use reasoning model DeepSeek-R1 instead of chat model DeepSeek-V3DeepSeek-V3 · HiPhO (13 Olympiads)gold medals: 1 -> 8Improved
use high reasoning effort for o4-minio4-mini · HiPhO (13 Olympiads)gold medals: 7 -> 6Improved
use Gemini-2.5-Flash as grader instead of GPT-4ograding pipeline · HiPhO step-level scoringagreement with human experts: under 1-point differencesImproved
compare best model to top human at IPhO-2025Gemini-2.5-Pro · IPhO-2025exam score: 29.2 -> 22.7Hurt
use open-source Kimi-K2-Instruct on PanMechanics-2025 versus top humanKimi-K2-Instruct · PanMechanics-2025exam score: 62.0 -> 65.9Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy