This paper presents an innovative approach to urban mobility simulation by integrating large-scale language models (LLMs) with agent-based modeling (ABM). Unlike traditional rule-based ABMs, the proposed framework leverages LLMs to enhance agent diversity and realism by generating synthetic population profiles, assigning routines and special places, and simulating personalized routes. Using real data, we simulate individual behaviors and large-scale mobility patterns in Taipei City. Key insights, such as route heatmaps and mode-specific metrics, provide urban planners with actionable information for policy decisions. Future work focuses on building a robust validation framework to ensure accuracy and reliability in urban planning applications.
