EgoCoT-Bench - Egocentric Grounding and Perception: leaderboard
EgoCoT-Bench (2026) Egocentric Grounding and Perception group (active object grounding, hand-object association and manipulation state perception) 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), Egocentric Grounding and Perception group (active object grounding, hand-object association and manipulation state perception); 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 397B A17B | 72.39 | #264 |
| 2 | Qwen 3.5 122B A10B | 72.39 | #346 |
| 3 | Qwen 3.5 Plus | 72.39 | #242 |
| 4 | Qwen 3.5 27B | 71.85 | #375 |
| 5 | Qwen 3 VL 8B | 71.43 | #917 |
| 6 | Qwen 3 VL 32B | 70.38 | #622 |
| 7 | Qwen 3 VL Plus | 70.24 | #512 |
| 8 | Qwen 3 VL 235B A22B | 68.63 | #485 |
| 9 | Qwen 3 VL 30B A3B | 67.88 | #784 |
| 10 | GPT-5.1 | 67.69 | #240 |
| 11 | GPT-5.2 | 66.62 | #185 |
| 12 | InternVL3.5-4B | 63 | #1930 |
| 13 | InternVL3.5-14B | 61.66 | #1504 |
| 14 | InternVL3.5-8B | 61.26 | #1604 |
| 15 | InternVL3.5-2B | 60.86 | #3284 |
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-egocentric-grounding-and-perception · How It Works · Data refreshed daily, snapshot 2026-11-21.