RobotEQ-Video - Synthetic Videos: leaderboard
The synthetic subset of RobotEQ-Video (Che, Lian et al., 2026): 1,218 generated egocentric videos built from a hierarchical world-state taxonomy, with 11,225 candidate robot actions to judge proper or improper. The score is the mean of accuracy, macro-F1, probability of agreement and consensus-weighted agreement with the annotators, in percent; higher is better.
Metric: Synthetic-subset Avg (%; action-properness judgment: given an egocentric robot-view video and a numbered list of candidate robot actions, the model labels each action proper or improper for socially appropriate proactive assistance; score is the mean of accuracy and macro-F1 against the majority-voted human label and of probability of agreement and consensus-weighted agreement with the four individual annotators; 1,218 generated videos, 11,225 actions). Source: arxiv.org. Saturation forecast: Around February 2027. 48 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 79.8 | #48 |
| 2 | Gemini 3.5 Flash | 79.7 | #98 |
| 3 | Seed 2.0 Lite | 79.4 | #286 |
| 4 | GPT-5.6 Luna | 79.1 | #117 |
| 5 | Gemini 3.1 Flash Lite | 77.8 | #363 |
| 6 | Qwen 3 VL 30B A3B Instruct | 76.7 | #870 |
| 7 | Qwen 3.5 27B (Thinking) | 76.7 | #358 (Qwen 3.5 27B) |
| 8 | Qwen 3.6 Plus | 76.2 | #217 |
| 9 | Gemma 4 12B (IT) | 76.1 | #688 |
| 10 | Gemma 4 31B (IT) | 76 | #343 |
| 11 | Kimi K2.6 | 75 | #173 |
| 12 | Claude Sonnet 5 | 75 | #108 |
| 13 | GLM-5V Turbo | 75 | #342 |
| 14 | Qwen 3.5 9B (Thinking) | 74.9 | #803 (Qwen 3.5 9B) |
| 15 | Qwen 3 VL 30B A3B (Thinking) | 74.5 | #710 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=roboteq-video-synthetic-videos · How It Works · Data refreshed daily, snapshot 2026-11-12.