The Internet has facilitated the rise of online education, which has become increasingly common. However, it has an obvious drawback: it is mainly a one-way channel of information transmission, with unilateral lectures by teachers and passive participation by students, and lacks interactivity. In this study, we propose an interactive online education framework that utilizes large language models for online programming education. The framework enables interactive learning and addresses the issue of passive participation by students. It uses educational large language models to provide extended knowledge and explanations during challenging instructional moments. It also uses the large language model to provide programming assistance by giving step-by-step coding hints and interactive guidance during the student’s online programming learning process. Finally, it dynamically combines course learning paths based on each student’s individual learning goals, resulting in an adaptive learning approach. This research offers valuable insights into the utilization of large language models to improve interactivity in online education, making a significant contribution to interactive teaching methods in the field of digital learning.