Preprint · 2024

VisualAgentBench: Towards Large Multimodal Models as Visual Foundation Agents

Xiao Liu, Tianjie Zhang, Yu Gu, Iat Long Iong, Yifan Xu, Xixuan Song, Shudan Zhang, Hanyu Lai, Xinyi Liu, Hanlin Zhao, Jiadai Sun, Xinyue Yang, Yu Yang, Zehan Qi, Shuntian Yao, Xueqiao Sun, Siyi Cheng, Qinkai Zheng, Hao Yu, Hanchen Zhang, Wenyi Hong, Ming Ding, Lihang Pan, Xiaotao Gu, Aohan Zeng, Zhengxiao Du, Chan Hee Song, Yu Su, Yuxiao Dong, Jie Tang

Abstract

Large Multimodal Models (LMMs) have ushered in a new era in artificial intelligence, merging capabilities in both language and vision to form highly capable Visual Foundation Agents. These agents are postulated to excel across a myriad of tasks, potentially approaching general artificial intelligence. However, existing benchmarks fail to sufficiently challenge or showcase the full potential of LMMs in complex, real-world environments. To address this gap, we introduce VisualAgentBench (VAB), a comprehensive and pioneering benchmark specifically designed to train and evaluate LMMs as visual foundation agents across diverse scenarios, including Embodied, Graphical User Interface, and Visual Design, with tasks formulated to probe the depth of LMMs' understanding and interaction capabilities. Through rigorous testing across nine proprietary LMM APIs and eight open models, we demonstrate the considerable yet still developing agent capabilities of these models. Additionally, VAB constructs a trajectory training set constructed through hybrid methods including Program-based Solvers, LMM Agent Bootstrapping, and Human Demonstrations, promoting substantial performance improvements in LMMs through behavior cloning. Our work not only aims to benchmark existing models but also provides a solid foundation for future development into visual foundation agents. Code, train \& test data, and part of fine-tuned open LMMs are available at \url{https://github.com/THUDM/VisualAgentBench}.

BibTeX Copy

@misc{liu_visualagentbench_2024,
	title = {{VisualAgentBench}: {Towards} {Large} {Multimodal} {Models} as {Visual} {Foundation} {Agents}},
	shorttitle = {{VisualAgentBench}},
	url = {http://arxiv.org/abs/2408.06327},
	doi = {10.48550/arXiv.2408.06327},
	abstract = {Large Multimodal Models (LMMs) have ushered in a new era in artificial intelligence, merging capabilities in both language and vision to form highly capable Visual Foundation Agents. These agents are postulated to excel across a myriad of tasks, potentially approaching general artificial intelligence. However, existing benchmarks fail to sufficiently challenge or showcase the full potential of LMMs in complex, real-world environments. To address this gap, we introduce VisualAgentBench (VAB), a comprehensive and pioneering benchmark specifically designed to train and evaluate LMMs as visual foundation agents across diverse scenarios, including Embodied, Graphical User Interface, and Visual Design, with tasks formulated to probe the depth of LMMs' understanding and interaction capabilities. Through rigorous testing across nine proprietary LMM APIs and eight open models, we demonstrate the considerable yet still developing agent capabilities of these models. Additionally, VAB constructs a trajectory training set constructed through hybrid methods including Program-based Solvers, LMM Agent Bootstrapping, and Human Demonstrations, promoting substantial performance improvements in LMMs through behavior cloning. Our work not only aims to benchmark existing models but also provides a solid foundation for future development into visual foundation agents. Code, train {\textbackslash}\& test data, and part of fine-tuned open LMMs are available at {\textbackslash}url\{https://github.com/THUDM/VisualAgentBench\}.},
	urldate = {2026-11-06},
	publisher = {arXiv},
	author = {Liu, Xiao and Zhang, Tianjie and Gu, Yu and Iong, Iat Long and Xu, Yifan and Song, Xixuan and Zhang, Shudan and Lai, Hanyu and Liu, Xinyi and Zhao, Hanlin and Sun, Jiadai and Yang, Xinyue and Yang, Yu and Qi, Zehan and Yao, Shuntian and Sun, Xueqiao and Cheng, Siyi and Zheng, Qinkai and Yu, Hao and Zhang, Hanchen and Hong, Wenyi and Ding, Ming and Pan, Lihang and Gu, Xiaotao and Zeng, Aohan and Du, Zhengxiao and Song, Chan Hee and Su, Yu and Dong, Yuxiao and Tang, Jie},
	month = aug,
	year = {2024},
	note = {arXiv:2408.06327 [cs.AI]},
	keywords = {Computer Science - Artificial Intelligence, Computer Science - Computation and Language, Computer Science - Computer Vision and Pattern Recognition},
}