EgoMemReason - Temporal Counting: leaderboard

EgoMemReason task (Wang et al., COLM 2026) that asks how many distinct instances of an object category have appeared in week-long egocentric video up to the question time, which requires re-identifying the same items across days. Accuracy in percent on 100 multiple-choice questions; higher is better.

Metric: Accuracy (%; 100 multiple-choice questions). Source: arxiv.org. Saturation forecast: Around 2032. 13 models tracked.

Top models

#ModelScoreOverall rank
1Molmo2-8B50#1488
2Gemini 3 Flash28#159
3Gemini 3.1 Pro (Preview)26#91

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

Interactive version: theaggregate.ai/benchmark?slug=egomemreason-temporal-counting · How It Works · Data refreshed daily, snapshot 2026-11-12.