Scientists' First Exam: Probing Cognitive Abilities of MLLM via Perception, Understanding, and Reasoning

TL;DR

The paper introduces the Scientists' First Exam (SFE), a comprehensive benchmark to evaluate the perception, understanding, and reasoning abilities of scientific Multimodal Large Language Models (MLLMs), revealing current models' limitations.

Contribution

It presents the SFE benchmark with 830 expert-verified VQA pairs across multiple scientific disciplines, addressing the gap in assessing MLLMs' scientific perception and reasoning capabilities.

Findings

Current models achieve only around 30% accuracy on SFE.

SFE covers 66 multimodal tasks across five scientific disciplines.

The benchmark highlights significant room for improvement in scientific MLLMs.

Abstract

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.