EgoMemReason - Activity Pattern Inference: leaderboard
EgoMemReason task (Wang et al., COLM 2026) that asks for the camera wearer's most likely next location or activity from routines seen across week-long egocentric video, for example where they go after lunch. Accuracy in percent on 50 multiple-choice questions; higher is better.
Metric: Accuracy (%; 50 multiple-choice questions). Source: arxiv.org. Saturation forecast: Around April 2027. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 48 | #91 |
| 2 | Gemini 3 Flash | 44 | #159 |
| 3 | Molmo2-8B | 22 | #1488 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=egomemreason-activity-pattern-inference · How It Works · Data refreshed daily, snapshot 2026-11-12.