Can MLLMs Reason Beyond Language? VisReason: A Comprehensive Benchmark for Vision-Centric Reasoning
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866913161178775552 |
|---|---|
| author | Guo, Longteng Wang, Yifan Huo, Pengkang Chen, Tailai Wu, Yuze Liu, Jing Zhu, Xinxin |
| author_facet | Guo, Longteng Wang, Yifan Huo, Pengkang Chen, Tailai Wu, Yuze Liu, Jing Zhu, Xinxin |
| contents | Recent multimodal large language models (MLLMs) achieve strong performance on visual reasoning benchmarks, yet it remains unclear to what extent such performance reflects reasoning directly grounded in visual evidence. We introduce VisReason, a benchmark for vision-centric reasoning in everyday scenarios where perception and inference are tightly coupled. VisReason contains 1,505 questions across 10 categories spanning perceptual, structural, and conceptual reasoning. Our evaluation shows that VisReason poses a qualitatively different challenge from existing benchmarks, exposing substantial gaps between humans and current MLLMs and revealing limited benefits from test-time reasoning strategies. VisReason offers a focused diagnostic for evaluating vision-centric reasoning beyond language. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_25364 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Can MLLMs Reason Beyond Language? VisReason: A Comprehensive Benchmark for Vision-Centric Reasoning Guo, Longteng Wang, Yifan Huo, Pengkang Chen, Tailai Wu, Yuze Liu, Jing Zhu, Xinxin Computer Vision and Pattern Recognition Recent multimodal large language models (MLLMs) achieve strong performance on visual reasoning benchmarks, yet it remains unclear to what extent such performance reflects reasoning directly grounded in visual evidence. We introduce VisReason, a benchmark for vision-centric reasoning in everyday scenarios where perception and inference are tightly coupled. VisReason contains 1,505 questions across 10 categories spanning perceptual, structural, and conceptual reasoning. Our evaluation shows that VisReason poses a qualitatively different challenge from existing benchmarks, exposing substantial gaps between humans and current MLLMs and revealing limited benefits from test-time reasoning strategies. VisReason offers a focused diagnostic for evaluating vision-centric reasoning beyond language. |
| title | Can MLLMs Reason Beyond Language? VisReason: A Comprehensive Benchmark for Vision-Centric Reasoning |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2605.25364 |