Smart education is a combination of AI and education, and its benefits are delayed. Existing research lacks an understanding of smart education continuance intention in the context of this delayed benefit. Thus, this study aimed to examine users' continuance intention in a delayed benefit context by integrating sensory stimulus, the extended UTAUT, and flow theory. The study focused on a mobile-based smart education (MBSE) system (Liulishuo) and collected data on 494 MBSE users. The PLS-SEM analysis suggested that the model is largely supported. Identifying MBSE's technical attributes as sensory stimuli, the extended UTAUT as primary information processing, flow as secondary information processing, and continuance intention as a decision, this study verifies the relationships between various constructs and their influence on MBSE continuance intention. The results also indicate that a user's flow state plays a crucial role in strengthening continuance intention and realizing the delayed benefits of smart education. Thus, this study provides a deeper understanding of continuous usage mechanisms as they relate to smart education.