metadata
license: mit
configs:
- config_name: ranking_panel
data_dir: rank
default: true
- config_name: missing_panel
data_dir: missing_panel
ComicVQA: A Benchmark for Visual Reasoning in Multimodal LLMs
Check out our paper here
ComicVQA is a specialized diagnostic benchmark designed to assess the visual reasoning capabilities of Multimodal Large Language Models (MLLMs) in the comics domain. ComicVQA focuses of 2 distinct reasoning tasks:
- Missing Panel Prediction: Predict the missing panel given a comic sequence.
- Panel Sorting: Choose which of the 4 comic sequences is the correct arrangement.
These tasks leverage the sequential nature of comics to expose vulnerabilities in how models synthesize visual and textual information over a temporal sequence.
Do refer to our Github respository for the handling of data and evaluation codes.
🖋️ Citation
Cite us!
@inproceedings{gan-etal-2026-comicvqa,
title = "{C}omic{VQA}: A Benchmark for Visual Reasoning in Multimodal {LLM}s",
author = "Gan, Esther and
Brown, Hannah and
Herel, David and
Kawaguchi, Kenji and
Kan, Min-Yen and
Shieh, Michael Qizhe",
editor = "Liakata, Maria and
Moreira, Viviane P. and
Zhang, Jiajun and
Jurgens, David",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
month = jul,
year = "2026",
address = "San Diego, California, United States",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-acl.1268/",
pages = "25347--25370",
ISBN = "979-8-89176-395-1",
abstract = "We introduce Comic Visual Question Answering (\textbf{ComicVQA}), a comics-based benchmark for evaluating MLLMs on visual reasoning. ComicVQA comprises of (i) \textbf{Missing Panel Prediction}, testing fine-grained visual grounding and (ii) \textbf{Panel Sorting}, which evaluates sequential narrative understanding. Proprietary models achieve up to 62.6{\%} on Missing Panel Prediction and 46.4{\%} on Panel Sorting, whereas open-source models reach only 47.7{\%} and 26.9{\%}, respectively. In contrast, human annotators achieve over 83{\%} accuracy on both tasks, revealing a large gap between current models and human-level multimodal understanding in comics. Through controlled ordering ablations and a detailed error taxonomy, we show that current MLLMs rely primarily on coarse temporal cues and struggle with fine-grained visual reasoning. These findings demonstrate ComicVQA as a diagnostic benchmark for advancing multimodal visual reasoning in comics."
}