Mind's Eye - Hierarchical Pattern Equivalence: 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 identifying the panel that violates the hierarchical structure shared by the other patterns; higher is better.

Metric: Accuracy (%) on the hierarchical pattern equivalence 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
1MindsEye InternVL3-8B (checkpoint unspecified)29.1
2Phi-3.5-Vision Instruct (MindsEye checkpoint unspecified)29.1
3Llama 3.2 11B Vision29#4264
4GPT-4o26.6#751
5InstructBLIP-4B (MindsEye checkpoint unspecified)26.4
6Qwen 2.5 VL 3B Instruct26.2#2713
7GT-Thinker-7B25.5
8Idefics-8B (MindsEye checkpoint unspecified)25.1
9Qwen 2.5 VL 7B Instruct24.2#1902
10Vision-G1-7B24.2
11llava-v1.6-mistral-7B23.7#3999
12BLIP-2-OPT-2.7B22.7#6196
13V-Thinker-8B (MindsEye checkpoint unspecified)22.6
14O322.4#237
15SmolVLM-2.2B (MindsEye checkpoint unspecified)21.2

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-hierarchical-pattern-equivalence · How It Works · Data refreshed daily, snapshot 2026-11-21.