UAV-DualCog - Landmark-Driven Action Decision: leaderboard
Metric: Answer accuracy (%; 1,024 questions asking which way the UAV should move to approach a target landmark; structured JSON answers on renders of simulated AerialVLN scenes, the discrete answer scored whatever the predicted box or interval; instant mode with explicit thinking disabled where a model allows it; higher is better). Source: arxiv.org. Saturation forecast: Around 2031. 36 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.3 Instant | 62.3 |
| 2 | Claude Sonnet 4.6 | 61 |
| 3 | Qwen 3.5 35B A3B (Non-reasoning) | 57.5 |
| 4 | GPT-5.5 (Non-reasoning) | 53.6 |
| 5 | Qwen 3.5 9B (Non-reasoning) | 52.2 |
| 6 | Qwen 3.5 27B (Non-reasoning) | 50.6 |
| 7 | Qwen 3.5 122B A10B (Non-reasoning) | 49 |
| 8 | Qwen 3.5 Plus (Non-reasoning) | 48.9 |
| 9 | Qwen 3.5 4B (Non-reasoning) | 47.5 |
| 10 | Qwen 3.5 397B A17B (Non-reasoning) | 46.2 |
| 11 | GLM-4.6V (Non-reasoning) | 44.2 |
| 12 | Qwen 3.7 Plus (Non-reasoning) | 43.4 |
| 13 | Qwen 3.6 Plus (Non-reasoning) | 42.5 |
| 14 | MiMo-V2-Omni | 38.2 |
| 15 | GPT-5.4 (Non-reasoning) | 36 |
Interactive version: theaggregate.ai/benchmark?slug=uav-dualcog-landmark-driven-action-decision · How It Works · Data refreshed daily, snapshot 2026-09-29.