Trip Planning on Natural-plan Trip Planning (test)
41.63EM AccuracyPlanGEN
Evaluation Results
| Method | Links | |
|---|---|---|
| PlanGENStrategy=Best of N, Backbone=Gemini-1.5-Pro2026.05 | 41.63 | |
| Gemini-1.5-Pro2026.05 | 34.75 | |
| RunAgent2026.05 | 14.73 | |
| Agent Workflow MemoryLLM Backbone=DeepSeek-V32026.06 | 7.3 | |
| GPT-4oMode=Plan Implementation2026.05 | 6.69 | |
| InducedLLM Backbone=DeepSeek-V32026.06 | 5.2 | |
| Hand-designedLLM Backbone=DeepSeek-V32026.06 | 4.2 | |
| GPT-4o2026.05 | 3.07 | |
| ReActLLM Backbone=DeepSeek-V32026.06 | 2.1 | |
| Program-of-ThoughtsLLM Backbone=DeepSeek-V32026.06 | 2.1 | |
| Chain-of-ThoughtLLM Backbone=DeepSeek-V32026.06 | 0 |