CoreCognition Image only Mechanicalreasoning: leaderboard

CoreCognition Image only 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: Rough model projection: around 2026. 77 models tracked. Updated 16 Nov 2026.

Top models

#ModelScoreOverall rank
1NVLM-D-72B77.08#2181
2Ovis1.6-Gemma2-27B74.62#1833
3Step-1.5V-mini74.15#2720
4step-1o-turbo-vision73.96
5Molmo-72B-092472.35#5272
6360VL-70B69.79#5987
7GLM-4V-Plus66.57#1400
8step-1v-32k60.32
9Cambrian-34B59.47#5295
10wemm-chat-2k-cn57.77
11wemm-chat-cn57.67
12Reka Core57.58#3349
13GLM-4V-9B57.39#3333
14WeMM56.44#5473
15omchat55.97

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