DisasterBench: leaderboard
Metric: Exact-match Accuracy (self-reported). Source: benchmarklist.com. 12 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 73.39 |
| 2 | Gemma 4 31B | 66.52 |
| 3 | Qwen 3 Max (Thinking) | 66.52 |
| 4 | GPT-5.4 | 65.67 |
| 5 | Qwen 3.5 9B | 65.67 |
| 6 | Qwen 3 Max | 63.95 |
| 7 | DeepSeek R1 | 62.66 |
| 8 | Qwen 3.5 27B | 62.23 |
| 9 | DeepSeek V3.2 | 61.8 |
| 10 | Llama 3.3 70B | 58.8 |
| 11 | Llama 3.1 8B | 28.33 |
Interactive version: theaggregate.ai/benchmark?slug=disasterbench · How It Works · Data refreshed daily, snapshot 2026-09-05.