RobotEQ-Video: leaderboard
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 December 2027. 48 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.6 Luna | 79.5 |
| 2 | GPT-5.5 | 79.3 |
| 3 | Gemini 3.5 Flash | 78.9 |
| 4 | Seed 2.0 Lite | 78.8 |
| 5 | Gemini 3.1 Flash Lite | 77.4 |
| 6 | Qwen 3.5 27B (Thinking) | 77.1 |
| 7 | Qwen 3 VL 30B A3B Instruct | 76 |
| 8 | Gemma 4 12B (IT) | 75.7 |
| 9 | Gemma 4 31B (IT) | 75.2 |
| 10 | Kimi K2.6 | 75.2 |
| 11 | Qwen 3.6 Plus | 74.9 |
| 12 | Qwen 3.5 9B (Thinking) | 74.5 |
| 13 | Qwen 3 VL 30B A3B (Thinking) | 74.4 |
| 14 | Qwen 3 VL 8B (Thinking) | 74.3 |
| 15 | GLM-4.1V-9B (Thinking) | 74.1 |
Interactive version: theaggregate.ai/benchmark?slug=roboteq-video · How It Works · Data refreshed daily, snapshot 2026-09-26.