DisasterBench - ReAct: leaderboard

ReAct planning, with reasoning interleaved with tool selection, 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: reasoning interleaved with tool selection over the current workflow state (temperature 0); 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)69.96#102
2Qwen 3.5 9B65.67#838
3GPT-5.464.38#138
4Gemma 4 31B62.23#390
5Qwen 3 Max62.23#420
6DeepSeek V3.261.8#425
7Qwen 3.5 27B59.23#380
8DeepSeek R156.22#484
9Ministral 3 14B50.64#1885
10Ministral 8B (DisasterBench checkpoint unspecified)44.21
11Llama 3.3 70B38.63#1523
12Ministral 3B (DisasterBench checkpoint unspecified)36.91
13Llama 3.1 8B26.61#5193

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=disasterbench-react · How It Works · Data refreshed daily, snapshot 2026-11-22.