Embodied Planning on ALFWorld Text N=134 unseen tasks Misaligned
501,049Token CountNL-PDDL
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| NL-PDDL2026.07 | 501,049 | 91 | |
| EvoPlanBackbone=Qwen3-32B2026.07 | 1,055,341 | 98.5 | |
| Gemini-2.0 FlashPlanner Type=Direct LLM/VLM-based, Backbone=Gemini-2.0 Flash2026.07 | 1,251,533 | 15 | |
| LLaMA-3.1Planner Type=Direct LLM/VLM-based, Backbone=LLaMA-3.12026.07 | 1,310,664 | 15 | |
| GPT-4oPlanner Type=Direct LLM/VLM-based, Backbone=GPT-4o2026.07 | 1,440,470 | 17 | |
| ReActPlanner Type=Reflective LLM-based, In-context Examples=with model2026.07 | 4,428,914 | 23 | |
| ReActPlanner Type=Reflective LLM-based, In-context Examples=with examples2026.07 | 5,215,612 | 79 |