RobotEQ-Video: leaderboard
RobotEQ-Video (Che, Lian et al., 2026) tests social proactive intelligence from a robot viewpoint: given an egocentric video and a list of candidate robot actions, the model judges which actions are proper or improper, against labels from four human annotators per video. The score is the mean of accuracy, macro-F1, probability of agreement and consensus-weighted agreement in percent, averaged over the synthetic and real video subsets; higher is better.
Metric: A-Avg: mean of the synthetic-subset and real-subset averages (%; 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). Source: arxiv.org. Saturation forecast: Around February 2027. 48 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.6 Luna | 79.5 | #117 |
| 2 | GPT-5.5 | 79.3 | #48 |
| 3 | Gemini 3.5 Flash | 78.9 | #98 |
| 4 | Seed 2.0 Lite | 78.8 | #286 |
| 5 | Gemini 3.1 Flash Lite | 77.4 | #363 |
| 6 | Qwen 3.5 27B (Thinking) | 77.1 | #358 (Qwen 3.5 27B) |
| 7 | Qwen 3 VL 30B A3B Instruct | 76 | #870 |
| 8 | Gemma 4 12B (IT) | 75.7 | #688 |
| 9 | Gemma 4 31B (IT) | 75.2 | #343 |
| 10 | Kimi K2.6 | 75.2 | #173 |
| 11 | Qwen 3.6 Plus | 74.9 | #217 |
| 12 | Qwen 3.5 9B (Thinking) | 74.5 | #803 (Qwen 3.5 9B) |
| 13 | Qwen 3 VL 30B A3B (Thinking) | 74.4 | #710 |
| 14 | Qwen 3 VL 8B (Thinking) | 74.3 | #887 |
| 15 | GLM-4.1V-9B (Thinking) | 74.1 | #1061 (GLM-4.1V-9B) |
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 · How It Works · Data refreshed daily, snapshot 2026-11-12.