In the modern e-commerce sector, cost control and operational efficiency in reverse logistics are particularly critical. To minimize the costs of reverse logistics networks, this study proposes the establishment of both a single logistics center and separate reverse logistics centers. A cost model encompassing both forward and reverse logistics operations is developed. The reverse logistics center specializes in handling returned goods, categorizing them into three types: those that can be directly resold, those requiring factory repair, and those to be directly discarded. The objective of this study is to minimize the overall operational costs of the reverse logistics network. An algorithm that uses a large model for initialization, followed by crossover and mutation for final optimization, is applied to solve the model, significantly enhancing the search efficiency and accuracy of the algorithm. The experimental results indicate that the algorithm optimized with the large model can effectively determine the optimal locations for logistics centers. Additionally, both the traditional single logistics center model and the separate reverse logistics center model demonstrate significant advantages in cost savings. Furthermore, the large model-optimized algorithm not only validates its effectiveness but also shows practical feasibility, providing an effective strategy for cost control and operational management in modern e-commerce reverse logistics.