DisasterBench (UAV) - Collapsed Building Analysis: leaderboard
Metric: Accuracy (%) on collapsed building analysis (CBA) 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 December 2026. 21 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 | 74.56 |
| 2 | Claude Sonnet 4.5 | 71.01 |
| 3 | Gemini 2.5 Pro | 70.41 |
| 4 | Gemini 2.5 Flash | 69.82 |
| 5 | Gemini 2.0 Flash | 65.68 |
| 6 | Grok 4 | 62.72 |
| 7 | InternVL3.5-8B | 55.62 |
| 8 | GPT-4o Mini | 55.03 |
| 9 | GPT-4o | 53.85 |
| 10 | InternVL2-8B | 42.6 |
| 11 | Qwen 2.5 VL 7B Instruct | 40.83 |
Interactive version: theaggregate.ai/benchmark?slug=disasterbench-uav-collapsed-building-analysis · How It Works · Data refreshed daily, snapshot 2026-09-29.