Mind's Eye - Dynamic Structural Correspondence: 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 matching structures that change in corresponding ways; higher is better.

Metric: Accuracy (%) on the dynamic structural correspondence 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
1llava-v1.6-mistral-7B32.4#3999
2Idefics-8B (MindsEye checkpoint unspecified)32.3
3Llama 3.2 11B Vision31.2#4264
4Qwen 2.5 VL 3B Instruct31#2713
5MindsEye InternVL3-8B (checkpoint unspecified)31
6Phi-3.5-Vision Instruct (MindsEye checkpoint unspecified)31
7Qwen 2.5 VL 7B Instruct30.4#1902
8GPT-4o30.3#751
9Gemini 2.5 Pro30#276
10Vision-G1-7B29.7
11GT-Thinker-7B28.1
12V-Thinker-8B (MindsEye checkpoint unspecified)27.1
13Qwen 2.5 VL 32B Instruct22.6#1062
14SmolVLM-2.2B (MindsEye checkpoint unspecified)19.2
15InstructBLIP-4B (MindsEye checkpoint unspecified)19.1

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-dynamic-structural-correspondence · How It Works · Data refreshed daily, snapshot 2026-11-21.