Abstract: Inventory scheduling plays a critical role in modern supply chains, with its primary objective being to balance production and demand through appropriate inventory allocation. However, as market demand becomes increasingly uncertain and product attributes more diverse, traditional experience-based inventory scheduling methods often struggle to effectively integrate product characteristics with demand fluctuations. Moreover, their decision outcomes lack interpretability, making it difficult to trace the rationale behind inventory schedules and reducing stakeholders' trust. To this end, we propose a novel multi-step reasoning framework for inventory scheduling based on large language models, termed InvLLM. Experimental evaluations on a large-scale real-world dataset provided by JD.com demonstrate that, compared to existing inventory scheduling methods, the proposed InvLLM framework increases service rate by 3.24% and improves overall inventory evaluation scores by 1.55%, while reducing inventory rate by 1.33%, thereby validating its effectiveness and practical applicability.