AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Paper Folding Puzzles: Can Multimodal Large Language Models Perform Spatial Reasoning?

Dibin Zhou, Yantao Xu, Zongming Huang, Zengwei Yan, Wenhao Liu, Yongwei Miao, Jianfeng Ren, Fuchang Liu

AAAI 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Chain-of-thought multimodal reasoning · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Introduces a paper-folding puzzle benchmark to test whether multimodal LLMs can perform spatial and abstract visual reasoning.

Problem

MLLMs lag human performance on abstract visual reasoning (AVR), and existing AVR benchmarks are limited to narrow, repetitive 2D patterns with simple spatial relations and few reasoning dimensions.

Approach

Uses paper-folding puzzles, inspired by human spatial reasoning, as a benchmark that requires inferring latent folding rules from visual question sets and generalizing them to novel scenarios, with richer spatial relationships than prior 2D AVR tasks.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
compare humans with best MLLM on spatial reasoning puzzlesMLLMs on PFP · PFP, five tasksaccuracy (%): human 72.64 vs best model 41.67 on single-step; 86.50 vs 41.00 on multi-stepHurt
increase folding complexity from 1-Fold to 3-Fold and corner-fold Others (ablation)MLLMs · PFP Single-Step Reasoningaccuracy (%): o4-mini 34.43 (1-Fold) -> 26.92 (Others)Hurt
add viewpoint variation to the 3D folding task (easy vs hard) (ablation)humans · PFP 3D Foldingaccuracy (%): 85.00 -> 80.33Hurt
add viewpoint variation to the 3D folding task (ablation)most MLLMs · PFP 3D Foldingaccuracy (%): o4-mini 30.67 -> 30.67No clear effect
change candidate net shapes in 2D unfolding (easy vs hard) (ablation)humans · PFP 2D Unfoldingaccuracy (%): 84.00 -> 83.33No clear effect
use closed-source reasoning model o4-mini instead of open-source modelsMLLMs · PFP overallaccuracy (%): Multi-step 41.00 vs open-source best 27.00Improved
scale up Qwen2.5-VL from 7B to 72B (ablation)Qwen2.5-VL · PFP Single-Step Reasoningaccuracy (%): 22.83 -> 31.67Improved
scale up Qwen2.5-VL from 32B to 72B (ablation)Qwen2.5-VL · PFP Single-Step Reasoningaccuracy (%): 29.17 -> 31.67No clear effect
use thinking or reasoning models versus non-reasoning GPT-4o on 3D tasksMLLMs · PFP 3D Folding and 2D Unfoldingaccuracy (%): 3D Folding o4-mini 30.33 vs GPT-4o 26.50No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy