This study analyzed the runoff reduction effects of green infrastructure (GI) and developed a physical and empirical hybrid model that can efficiently evaluate these effects. Focusing on the flood-prone Seocho District, a Storm Water Management Model (SWMM)-based Long Short-Term Memory (LSTM) model was developed using various low-impact development (LID) facility installation scenarios involving green roofs, permeable pavements, and rain gardens. The results showed that the model enhanced the efficiency and accuracy of simulations incorporating green infrastructure. The key findings of this study are as follows: 1) Scenarios with green infrastructure effectively responded to both short- and long-term changes in rainfall patterns. 2) The SWMM-based LSTM model predicted runoff significantly faster and more efficiently than the conventional SWMM simulation, reducing computation time from 1 h and 48 min to just 6.08 s when applied to 2023 rainfall data. These results indicate that utilizing an SWMM-based LSTM model when introducing green infrastructure can provide more rapid and accurate runoff predictions, which is highly beneficial for real-time flood management and emergency response. However, this study has some limitations. The analysis focused on a single scenario and a single district, which may limit the generalizability of the findings. Future research should evaluate multiple LID configurations and expand regional applicability to improve the model’s universality.