CoreCognition Single Image Mechanicalreasoning: leaderboard

CoreCognition Single Image Mechanicalreasoning — CoreCognition native component accuracy and reported mean. Image+Video, image-only and single-image populations are separate; no pooling or replacement across cohorts. Metric: Accuracy (%). Source: https://grow-ai-like-a-child.github.io/CoreCognition/leaderboard

Metric: Accuracy (%). Source: grow-ai-like-a-child.github.io. Saturation forecast: Estimated already saturated. 33 models tracked. Updated 16 Nov 2026.

Top models

#ModelSettingScoreOverall rank
1qwen2.5-vl-72b-instruct_videoCoreCognition run: native-setting=qwen2.5-vl-72b-instruct_video85.66
2LLaVA-Video-72B-Qwen277.42#1228
3InternVL2.5-38B70.79#1413
4VideoLLaMA2-72B69.71#1355
5InternVL2.5-78B67.56#1044
6Llama 3.2 90B Vision Instruct67.56#1769
7llava-next-video-32B-qwen67.28
8LLaVA-Video-7B-Qwen263.68#1854
9qwen2.5-vl-3b-instruct_videoCoreCognition run: native-setting=qwen2.5-vl-3b-instruct_video63.17
10InternVL2.5-8B63.08#2444
11qwen2.5-vl-7b-instruct_videoCoreCognition run: native-setting=qwen2.5-vl-7b-instruct_video62.99
12LLaVA-CoT61.83#3271
13InternVL2.5-26B61.56#2269
14yi-vision-v260.39
15llava-video-7B-qwen2-video-only56.14

No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.

Interactive version: theaggregate.ai/benchmark?slug=corecognition-single-image-mechanicalreasoning · How It Works · Data refreshed daily, snapshot 2026-11-16.