RBench-V: A Primary Assessment for Visual Reasoning Models with Multi-modal Outputs

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Guo, Meng-Hao, Chu, Xuanyu, Yang, Qianrui, Mo, Zhe-Han, Shen, Yiqing, Li, Pei-lin, Lin, Xinjie, Zhang, Jinnian, Chen, Xin-Sheng, Zhang, Yi, Nakayama, Kiyohiro, Geng, Zhengyang, Peng, Houwen, Hu, Han, Hu, Shi-Min
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916754138071040
author Guo, Meng-Hao
Chu, Xuanyu
Yang, Qianrui
Mo, Zhe-Han
Shen, Yiqing
Li, Pei-lin
Lin, Xinjie
Zhang, Jinnian
Chen, Xin-Sheng
Zhang, Yi
Nakayama, Kiyohiro
Geng, Zhengyang
Peng, Houwen
Hu, Han
Hu, Shi-Min
author_facet Guo, Meng-Hao
Chu, Xuanyu
Yang, Qianrui
Mo, Zhe-Han
Shen, Yiqing
Li, Pei-lin
Lin, Xinjie
Zhang, Jinnian
Chen, Xin-Sheng
Zhang, Yi
Nakayama, Kiyohiro
Geng, Zhengyang
Peng, Houwen
Hu, Han
Hu, Shi-Min
contents The rapid advancement of native multi-modal models and omni-models, exemplified by GPT-4o, Gemini, and o3, with their capability to process and generate content across modalities such as text and images, marks a significant milestone in the evolution of intelligence. Systematic evaluation of their multi-modal output capabilities in visual thinking processes (also known as multi-modal chain of thought, M-CoT) becomes critically important. However, existing benchmarks for evaluating multi-modal models primarily focus on assessing multi-modal inputs and text-only reasoning while neglecting the importance of reasoning through multi-modal outputs. In this paper, we present a benchmark, dubbed RBench-V, designed to assess models' vision-indispensable reasoning abilities. To construct RBench-V, we carefully hand-pick 803 questions covering math, physics, counting, and games. Unlike previous benchmarks that typically specify certain input modalities, RBench-V presents problems centered on multi-modal outputs, which require image manipulation such as generating novel images and constructing auxiliary lines to support the reasoning process. We evaluate numerous open- and closed-source models on RBench-V, including o3, Gemini 2.5 Pro, Qwen2.5-VL, etc. Even the best-performing model, o3, achieves only 25.8% accuracy on RBench-V, far below the human score of 82.3%, highlighting that current models struggle to leverage multi-modal reasoning. Data and code are available at https://evalmodels.github.io/rbenchv
format Preprint
id arxiv_https___arxiv_org_abs_2505_16770
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle RBench-V: A Primary Assessment for Visual Reasoning Models with Multi-modal Outputs
Guo, Meng-Hao
Chu, Xuanyu
Yang, Qianrui
Mo, Zhe-Han
Shen, Yiqing
Li, Pei-lin
Lin, Xinjie
Zhang, Jinnian
Chen, Xin-Sheng
Zhang, Yi
Nakayama, Kiyohiro
Geng, Zhengyang
Peng, Houwen
Hu, Han
Hu, Shi-Min
Computer Vision and Pattern Recognition
The rapid advancement of native multi-modal models and omni-models, exemplified by GPT-4o, Gemini, and o3, with their capability to process and generate content across modalities such as text and images, marks a significant milestone in the evolution of intelligence. Systematic evaluation of their multi-modal output capabilities in visual thinking processes (also known as multi-modal chain of thought, M-CoT) becomes critically important. However, existing benchmarks for evaluating multi-modal models primarily focus on assessing multi-modal inputs and text-only reasoning while neglecting the importance of reasoning through multi-modal outputs. In this paper, we present a benchmark, dubbed RBench-V, designed to assess models' vision-indispensable reasoning abilities. To construct RBench-V, we carefully hand-pick 803 questions covering math, physics, counting, and games. Unlike previous benchmarks that typically specify certain input modalities, RBench-V presents problems centered on multi-modal outputs, which require image manipulation such as generating novel images and constructing auxiliary lines to support the reasoning process. We evaluate numerous open- and closed-source models on RBench-V, including o3, Gemini 2.5 Pro, Qwen2.5-VL, etc. Even the best-performing model, o3, achieves only 25.8% accuracy on RBench-V, far below the human score of 82.3%, highlighting that current models struggle to leverage multi-modal reasoning. Data and code are available at https://evalmodels.github.io/rbenchv
title RBench-V: A Primary Assessment for Visual Reasoning Models with Multi-modal Outputs
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2505.16770