CoreCognition Single Image Perceptualconstancy: leaderboard

CoreCognition Single Image Perceptualconstancy — 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_video84.69
2llava-next-video-32B-qwen72.38
3InternVL2.5-78B72.29#1044
4LLaVA-Video-7B-Qwen272.09#1854
5LLaVA-Video-72B-Qwen271.51#1228
6VideoLLaMA2-72B68.41#1355
7qwen2.5-vl-7b-instruct_videoCoreCognition run: native-setting=qwen2.5-vl-7b-instruct_video68.02
8yi-vision-v267.64
9LLaVA-CoT67.34#3271
10InternVL2.5-26B66.86#2269
11videollama2-7B-16f65.99
12VideoLLaMA2.1-7B-16F65.12#1924
13InternVL2.5-4B64.53#2583
14qwen2.5-vl-3b-instruct_videoCoreCognition run: native-setting=qwen2.5-vl-3b-instruct_video64.34
15Llama 3.2 90B Vision Instruct62.21#1769

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-perceptualconstancy · How It Works · Data refreshed daily, snapshot 2026-11-16.