DisasterBench - Tree-of-Thoughts: leaderboard

Tree-of-thoughts planning, a breadth-first search over candidate plans with LLM pruning, 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: breadth-first search over candidate plans with LLM pruning (temperature 0.5); higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 13 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3.1 Pro (Preview)73.39#99
2Qwen 3 Max63.95#415
3GPT-5.462.23#135
4Qwen 3.5 9B59.23#828
5Qwen 3.5 27B59.23#375
6Gemma 4 31B58.37#382
7DeepSeek R153.22#482
8DeepSeek V3.250.21#419
9Llama 3.3 70B41.63#1514
10Ministral 8B (DisasterBench checkpoint unspecified)21.03
11Ministral 3B (DisasterBench checkpoint unspecified)18.88
12Llama 3.1 8B16.31#5196
13Ministral 3 14B16.31#1870

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

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