Performance comparison of Procedural Graph against 7 baselines (Vanilla ReAct, MemoryBank, RAP, ExpeL, AutoGuide, AWM, KnowAgent) on ALFWorld (embodied household task completion on standard unseen split (134 test games)) using Claude Sonnet 4.6. Procedural Graph achieves 93.28% compared to the best baseline score of 91.79% (margin of +1.49%).

model: Claude Sonnet 4.6 · method: Vanilla ReAct (Yao et al., 2023b) · benchmark: ALFWorld

model: Claude Sonnet 4.6 · method: MemoryBank (Zhong et al., 2024) · benchmark: ALFWorld

model: Claude Sonnet 4.6 · method: RAP (Kagaya et al., 2024) · benchmark: ALFWorld

model: Claude Sonnet 4.6 · method: ExpeL (Zhao et al., 2024) · benchmark: ALFWorld

model: Claude Sonnet 4.6 · method: AutoGuide (Fu et al., 2024) · benchmark: ALFWorld

model: Claude Sonnet 4.6 · method: AWM (Wang et al., 2025b) · benchmark: ALFWorld

model: Claude Sonnet 4.6 · method: KnowAgent (Zhu et al., 2025) · benchmark: ALFWorld

model: Claude Sonnet 4.6 · method: Procedural Graph (Ours) · benchmark: ALFWorld