Mind's Eye - Paper Folding: leaderboard

Mind's Eye (2026) is a multiple-choice benchmark of visuospatial reasoning tasks adapted from human intelligence tests, with procedurally generated items answered with chain-of-thought prompting. This board is accuracy in percent on predicting hole positions after folding and punching paper; higher is better.

Metric: Accuracy (%) on the paper folding (fold and hole punch) task of Mind's Eye (4-option multiple choice, 100 procedurally generated items, chain-of-thought prompt, answers extracted by an LLM evaluator; keys uniformly distributed over the options); higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 18 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Idefics-8B (MindsEye checkpoint unspecified)41.5
2Qwen 2.5 VL 32B Instruct32#1062
3Gemini 2.5 Pro31.1#276
4V-Thinker-8B (MindsEye checkpoint unspecified)29.2
5GPT-4o29#751
6Vision-G1-7B28.1
7GT-Thinker-7B27.9
8PaliGemma-3B (MindsEye checkpoint unspecified)26.4
9SmolVLM-2.2B (MindsEye checkpoint unspecified)26
10O325.6#237
11Qwen 2.5 VL 7B Instruct25.1#1902
12Qwen 2.5 VL 3B Instruct25#2713
13MindsEye InternVL3-8B (checkpoint unspecified)24.6
14Llama 3.2 11B Vision24.5#4264
15Phi-3.5-Vision Instruct (MindsEye checkpoint unspecified)24.3

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=mind-s-eye-paper-folding · How It Works · Data refreshed daily, snapshot 2026-11-21.