EgoCoT-Bench - Spatio-Temporal Retrospection: leaderboard
Metric: Answer accuracy (%, four-way multiple choice, strict exact match) of zero-shot MLLMs on egocentric manipulation videos of EgoCoT-Bench (351 clips, 3,172 human-verified questions built from spatio-temporal scene graphs), Spatio-Temporal Retrospection group (state, location or visibility, and hand-object temporal retrospection); higher is better. Source: arxiv.org. Saturation forecast: Not forecast. 19 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.5 27B | 69.1 |
| 2 | Qwen 3.5 122B A10B | 65.11 |
| 3 | Qwen 3.5 Plus | 62.67 |
| 4 | Qwen 3.5 397B A17B | 62.55 |
| 5 | GPT-5.2 | 62.42 |
| 6 | Qwen 3 VL 235B A22B | 59.14 |
| 7 | GPT-5.1 | 55.23 |
Interactive version: theaggregate.ai/benchmark?slug=egocot-bench-spatio-temporal-retrospection · How It Works · Data refreshed daily, snapshot 2026-10-07.