MulRobBench: leaderboard
Metric: Semantic protocol-decision score (%; aggregate over constraint extraction, target disambiguation, safe action, abstention or reobservation and collaboration trigger; 3,024-sample strict evaluation set of real UAV observations with protocol constraints, degradations and operator language; controlled semantic scores in [0, 1] x100). Source: arxiv.org. Saturation forecast: Around 2032. 17 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.5 9B | 49.78 |
| 2 | Qwen 3.5 4B | 48.88 |
| 3 | Gemma 4 E4B | 46.29 |
| 4 | InternVL3.5-8B | 45.79 |
| 5 | Qwen 2.5 VL 7B | 44.99 |
| 6 | Gemma 4 E2B | 43.49 |
| 7 | Qwen 3.5 2B | 43.27 |
| 8 | aya-vision-8B | 38.61 |
Interactive version: theaggregate.ai/benchmark?slug=mulrobbench · How It Works · Data refreshed daily, snapshot 2026-09-29.