DisasterBench (UAV) - Risk Assessment and Evaluation: leaderboard
Metric: Accuracy (%) on pre-disaster risk assessment and evaluation (RAE) questions on the 3,000-question test split of low-altitude UAV disaster images (14 disaster scene types), multiple choice scored by strict exact match on the final option letter; higher is better. Source: arxiv.org. Saturation forecast: Around 2030. 21 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 | 66.67 |
| 2 | InternVL2-8B | 61.4 |
| 3 | Gemini 2.5 Pro | 58.77 |
| 4 | Qwen 2.5 VL 7B Instruct | 56.14 |
| 5 | Gemini 2.5 Flash | 55.26 |
| 6 | Grok 4 | 53.51 |
| 7 | Claude Sonnet 4.5 | 52.63 |
| 8 | Gemini 2.0 Flash | 47.37 |
| 9 | InternVL3.5-8B | 43.86 |
| 10 | GPT-4o | 40.35 |
| 11 | GPT-4o Mini | 27.19 |
Interactive version: theaggregate.ai/benchmark?slug=disasterbench-uav-risk-assessment-and-evaluation · How It Works · Data refreshed daily, snapshot 2026-09-29.