DisasterBench - Direct Planning: leaderboard

Direct planning, the whole plan generated in one pass, on DisasterBench (2026), which tests whether an LLM can compose executable disaster-response workflows from 26 specialized tools under typed interface constraints. Exact-match accuracy over the planning tasks (every step must match in tool, parameter bindings and dependencies), percent; higher is better.

Metric: Exact-match accuracy (%) over the 233 DisasterBench planning tasks (35 single-tool, 166 chain and 32 branching workflows over 26 disaster-response tools): a plan counts only when every step matches the ground truth in tool identity, parameter bindings and inter-step dependencies; planning method: the complete plan generated in one pass (temperature 0); higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 13 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Gemini 3.1 Pro (Preview)68.24#99
2Gemma 4 31B66.52#382
3Qwen 3.5 27B58.37#375
4DeepSeek V3.257.94#419
5Qwen 3 MaxThinking57.51#415
6GPT-5.454.08#135
7DeepSeek R154.08#482
8Qwen 3.5 9B45.92#828
9Llama 3.3 70B38.63#1514
10Ministral 8B (DisasterBench checkpoint unspecified)25.32
11Ministral 3 14B20.17#1870
12Llama 3.1 8B15.88#5196
13Ministral 3B (DisasterBench checkpoint unspecified)10.3

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=disasterbench-direct-planning · How It Works · Data refreshed daily, snapshot 2026-11-21.