EgoCoT-Bench - Spatio-Temporal Retrospection: leaderboard
EgoCoT-Bench (2026) Spatio-Temporal Retrospection group (state, location or visibility, and hand-object temporal retrospection) on egocentric manipulation videos. Four-way multiple choice accuracy in percent; higher is better.
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: Around February 2027. 19 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.5 27B | 69.1 | #375 |
| 2 | Qwen 3 VL 30B A3B | 65.49 | #784 |
| 3 | Qwen 3.5 122B A10B | 65.11 | #346 |
| 4 | InternVL3.5-2B | 64.18 | #3284 |
| 5 | InternVL3.5-8B | 63.3 | #1604 |
| 6 | Qwen 3.5 Plus | 62.67 | #242 |
| 7 | Qwen 3.5 397B A17B | 62.55 | #264 |
| 8 | GPT-5.2 | 62.42 | #185 |
| 9 | InternVL3.5-14B | 61.92 | #1504 |
| 10 | Qwen 3 VL 8B | 61.75 | #917 |
| 11 | Qwen 3 VL Plus | 59.75 | #512 |
| 12 | Qwen 3 VL 235B A22B | 59.14 | #485 |
| 13 | Qwen 3 VL 32B | 58.76 | #622 |
| 14 | LLaVA-OneVision-1.5-4B | 56.62 | #1855 |
| 15 | GPT-5.1 | 55.23 | #240 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=egocot-bench-spatio-temporal-retrospection · How It Works · Data refreshed daily, snapshot 2026-11-21.