MDK12-Bench: A Comprehensive Evaluation of Multimodal Large Language Models on Multidisciplinary Exams

TL;DR

MDK12-Bench is a large-scale, multidisciplinary benchmark designed to comprehensively evaluate multimodal large language models across various real-world exams, addressing limitations of existing benchmarks and promoting model robustness.

Contribution

Introduces MDK12-Bench, a comprehensive, real-world exam-based benchmark with a dynamic evaluation framework to better assess MLLMs' generalization and reasoning abilities.

Findings

Current MLLMs show significant limitations in multiple evaluation dimensions.

The benchmark reveals weaknesses in model robustness and generalization.

Knowledge-driven reasoning improves problem-solving performance.

Abstract

Multimodal large language models (MLLMs), which integrate language and visual cues for problem-solving, are crucial for advancing artificial general intelligence (AGI). However, current benchmarks for measuring the intelligence of MLLMs suffer from limited scale, narrow coverage, and unstructured knowledge, offering only static and undifferentiated evaluations. To bridge this gap, we introduce MDK12-Bench, a large-scale multidisciplinary benchmark built from real-world K-12 exams spanning six disciplines with 141K instances and 6,225 knowledge points organized in a six-layer taxonomy. Covering five question formats with difficulty and year annotations, it enables comprehensive evaluation to capture the extent to which MLLMs perform over four dimensions: 1) difficulty levels, 2) temporal (cross-year) shifts, 3) contextual shifts, and 4) knowledge-driven reasoning. We propose a novel dynamic evaluation framework that introduces unfamiliar visual, textual, and question form shifts to challenge model generalization while improving benchmark objectivity and longevity by mitigating data contamination. We further evaluate knowledge-point reference-augmented generation (KP-RAG) to examine the role of knowledge in problem-solving. Key findings reveal limitations in current MLLMs in multiple aspects and provide guidance for enhancing model robustness, interpretability, and AI-assisted education.