PlanningBench: leaderboard
Planning is a fundamental capability for large language models (LLMs) because such complex tasks require models to coordinate goals, constraints, resources, and long-term consequences into executable and verifiable solutions. Existing planning benchmarks, however, usually treat planning data as fixed collections of instances rather than controllable generation targets. This limits scenario coverag
Metric: All-pass (%) (self-reported). Source: benchmarklist.com. Saturation forecast: Around March 2027. 18 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.4 (xHigh) | 63.17 | #79 (GPT-5.4) |
| 2 | GPT-5.4 (High) | 58.56 | #79 (GPT-5.4) |
| 3 | GPT-5.4 (Medium) | 58.09 | #79 (GPT-5.4) |
| 4 | Gemini 3.1 Pro (Preview) | 53.25 | #58 |
| 5 | Seed 2.0 Pro (High) | 44.33 | #118 (Seed 2.0 Pro) |
| 6 | DeepSeek V3.2 (Thinking) | 37.13 | #205 (DeepSeek V3.2) |
| 7 | Hy3-preview | 36.17 | #157 |
| 8 | Qwen 3.5 Plus (Thinking) | 34.03 | #125 (Qwen 3.5 Plus) |
| 9 | DeepSeek R1 | 33.07 | #251 |
| 10 | Gemini 2.5 Pro | 31.7 | #152 |
| 11 | Seed 2.0 Pro (Medium) | 30.63 | #118 (Seed 2.0 Pro) |
| 12 | DeepSeek V3.2 Exp | 29.27 | #234 |
| 13 | Seed 1.8 | 22.4 | #139 |
| 14 | Qwen 3 32B | 0.27 | #443 |
| 15 | Qwen 3 8B | 0 | #674 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=planningbench · How It Works · Data refreshed daily, snapshot 2026-10-20.