Saved in:
Bibliographic Details
Main Authors: Zhou, Yuhao, Wang, Yiheng, He, Xuming, Shen, Ao, Xiao, Ruoyao, Li, Zhiwei, Feng, Qiantai, Guo, Zijie, Yang, Yuejin, Wu, Hao, Huang, Wenxuan, Wei, Jiaqi, Si, Dan, Yao, Xiuqi, Bu, Jia, Huang, Haiwen, Wang, Manning, Fu, Tianfan, Tang, Shixiang, Fei, Ben, Zhou, Dongzhan, Ling, Fenghua, Lu, Yan, Sun, Siqi, Li, Chenhui, Zheng, Guanjie, Lv, Jiancheng, Zhang, Wenlong, Bai, Lei
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2506.10521
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909901329006592
author Zhou, Yuhao
Wang, Yiheng
He, Xuming
Shen, Ao
Xiao, Ruoyao
Li, Zhiwei
Feng, Qiantai
Guo, Zijie
Yang, Yuejin
Wu, Hao
Huang, Wenxuan
Wei, Jiaqi
Si, Dan
Yao, Xiuqi
Bu, Jia
Huang, Haiwen
Wang, Manning
Fu, Tianfan
Tang, Shixiang
Fei, Ben
Zhou, Dongzhan
Ling, Fenghua
Lu, Yan
Sun, Siqi
Li, Chenhui
Zheng, Guanjie
Lv, Jiancheng
Zhang, Wenlong
Bai, Lei
author_facet Zhou, Yuhao
Wang, Yiheng
He, Xuming
Shen, Ao
Xiao, Ruoyao
Li, Zhiwei
Feng, Qiantai
Guo, Zijie
Yang, Yuejin
Wu, Hao
Huang, Wenxuan
Wei, Jiaqi
Si, Dan
Yao, Xiuqi
Bu, Jia
Huang, Haiwen
Wang, Manning
Fu, Tianfan
Tang, Shixiang
Fei, Ben
Zhou, Dongzhan
Ling, Fenghua
Lu, Yan
Sun, Siqi
Li, Chenhui
Zheng, Guanjie
Lv, Jiancheng
Zhang, Wenlong
Bai, Lei
contents Scientific discoveries increasingly rely on complex multimodal reasoning based on information-intensive scientific data and domain-specific expertise. Empowered by expert-level scientific benchmarks, scientific Multimodal Large Language Models (MLLMs) hold the potential to significantly enhance this discovery process in realistic workflows. However, current scientific benchmarks mostly focus on evaluating the knowledge understanding capabilities of MLLMs, leading to an inadequate assessment of their perception and reasoning abilities. To address this gap, we present the Scientists' First Exam (SFE) benchmark, designed to evaluate the scientific cognitive capacities of MLLMs through three interconnected levels: scientific signal perception, scientific attribute understanding, scientific comparative reasoning. Specifically, SFE comprises 830 expert-verified VQA pairs across three question types, spanning 66 multimodal tasks across five high-value disciplines. Extensive experiments reveal that current state-of-the-art GPT-o3 and InternVL-3 achieve only 34.08% and 26.52% on SFE, highlighting significant room for MLLMs to improve in scientific realms. We hope the insights obtained in SFE will facilitate further developments in AI-enhanced scientific discoveries.
format Preprint
id arxiv_https___arxiv_org_abs_2506_10521
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Scientists' First Exam: Probing Cognitive Abilities of MLLM via Perception, Understanding, and Reasoning
Zhou, Yuhao
Wang, Yiheng
He, Xuming
Shen, Ao
Xiao, Ruoyao
Li, Zhiwei
Feng, Qiantai
Guo, Zijie
Yang, Yuejin
Wu, Hao
Huang, Wenxuan
Wei, Jiaqi
Si, Dan
Yao, Xiuqi
Bu, Jia
Huang, Haiwen
Wang, Manning
Fu, Tianfan
Tang, Shixiang
Fei, Ben
Zhou, Dongzhan
Ling, Fenghua
Lu, Yan
Sun, Siqi
Li, Chenhui
Zheng, Guanjie
Lv, Jiancheng
Zhang, Wenlong
Bai, Lei
Artificial Intelligence
Computation and Language
Scientific discoveries increasingly rely on complex multimodal reasoning based on information-intensive scientific data and domain-specific expertise. Empowered by expert-level scientific benchmarks, scientific Multimodal Large Language Models (MLLMs) hold the potential to significantly enhance this discovery process in realistic workflows. However, current scientific benchmarks mostly focus on evaluating the knowledge understanding capabilities of MLLMs, leading to an inadequate assessment of their perception and reasoning abilities. To address this gap, we present the Scientists' First Exam (SFE) benchmark, designed to evaluate the scientific cognitive capacities of MLLMs through three interconnected levels: scientific signal perception, scientific attribute understanding, scientific comparative reasoning. Specifically, SFE comprises 830 expert-verified VQA pairs across three question types, spanning 66 multimodal tasks across five high-value disciplines. Extensive experiments reveal that current state-of-the-art GPT-o3 and InternVL-3 achieve only 34.08% and 26.52% on SFE, highlighting significant room for MLLMs to improve in scientific realms. We hope the insights obtained in SFE will facilitate further developments in AI-enhanced scientific discoveries.
title Scientists' First Exam: Probing Cognitive Abilities of MLLM via Perception, Understanding, and Reasoning
topic Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2506.10521