CoreCognition Image + Video Mechanicalreasoning: leaderboard

CoreCognition Image + Video 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. 120 models tracked. Updated 16 Nov 2026.

Top models

#ModelSettingScoreOverall rank
1Qwen-VL-Max-2025-01-2586.2
2Qwen 2.5 VL 72B Instruct85.66#848
3O1 (2024-12-17)85.13#313
4Qwen 2 VL 72B Instruct81.81#1388
5Gemini 1.5 Pro80.56#1061
6InternVL2-76B78.85#1519
7GPT-4o77.51#744
8llava-video-72b-qwen2_multi_frameCoreCognition run: native-setting=llava-video-72b-qwen2_multi_frame76.79
9GPT-4 Turbo75.81#999
10Gemini 1.5 Flash72.13#1694
11Gemma 3 27B (IT)70.53#1322
12Gemma 3 12B (IT)70.35#1884
13LLaVA-NeXT-72B70.07#3185
14InternVL2-4B69.89#4033
15InternVL2-8B69.53#3186

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

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