EgoCoT-Bench - Hand-Object Grounded CoT: 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), Hand-Object Grounded CoT subtask (answer through a reasoning chain grounded in hand-object and temporal cues); higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 19 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.2 | 84.03 |
| 2 | Qwen 3.5 397B A17B | 83.61 |
| 3 | Qwen 3.5 Plus | 82.77 |
| 4 | Qwen 3.5 27B | 80.67 |
| 5 | GPT-5.1 | 79.83 |
| 6 | Qwen 3.5 122B A10B | 79.41 |
| 7 | Qwen 3 VL 235B A22B | 78.99 |
Interactive version: theaggregate.ai/benchmark?slug=egocot-bench-hand-object-grounded-cot · How It Works · Data refreshed daily, snapshot 2026-10-07.