PlanBench-V - Decision Making: leaderboard

Metric: Mean score (out of 2) on choosing between planning alternatives under stated constraints, implementation level; answers to expert-curated questions on Chinese territorial spatial planning maps are scored 0 to 2 by a GPT-4o-mini judge (temperature 0) against structured reference answers with annotated critical points; objective items use exact match or semantic similarity; higher is better. Source: arxiv.org. Saturation forecast: Around December 2026. 17 models tracked.

Top models

#ModelScore
1Qwen 3.6 Plus1.71
2Gemini 2.5 Pro1.53
3GPT-5.41.51
4Qwen 3.6 Flash1.44
5GPT-4o1.43
6Claude Opus 4.71.32
7Kimi K2.61.32
8GPT-4o Mini1.18
9InternVL3-8B1.07
10Qwen 2.5 VL 7B Instruct1.05
11InternVL3-14B0.92
12Qwen 2 VL 2B0.79
13Qwen 2 VL 7B0.66

Interactive version: theaggregate.ai/benchmark?slug=planbench-v-decision-making · How It Works · Data refreshed daily, snapshot 2026-09-29.