ClassMeta is a GPT-4 powered virtual agent designed to promote classroom participation in virtual reality (VR) classrooms. The agent is embodied as a 3D avatar and interacts with instructors and students using both spoken language and body gestures. ClassMeta is designed to exert peer influence by displaying behaviors commonly observed among active students, such as taking notes, responding to instructors, correcting distracted students, and participating in discussions. The agent uses real-time contextual information from the classroom to adapt its behaviors and engage in dynamic interactions. ClassMeta supports classroom participation through various interactions, including asking questions, raising questions, reminding students of key points, and facilitating discussions. The agent's behaviors are designed to encourage active participation and create a positive behavioral norm in the classroom. The design of ClassMeta leverages the capabilities of large language models (LLMs) to generate contextually coherent responses and simulate diverse human behaviors. The agent's role is configured to resemble that of a typical student with limited knowledge, rather than an all-knowing expert, to ensure its behaviors are realistic and conducive to classroom dynamics. A comparative user study was conducted to evaluate the effectiveness of ClassMeta in improving classroom participation. The study found that students in the ClassMeta condition showed increased attention to the agent, better note-taking quality, and more active participation in discussions compared to the baseline condition. The results suggest that ClassMeta can effectively promote classroom participation and enhance the learning experience in VR classrooms.ClassMeta is a GPT-4 powered virtual agent designed to promote classroom participation in virtual reality (VR) classrooms. The agent is embodied as a 3D avatar and interacts with instructors and students using both spoken language and body gestures. ClassMeta is designed to exert peer influence by displaying behaviors commonly observed among active students, such as taking notes, responding to instructors, correcting distracted students, and participating in discussions. The agent uses real-time contextual information from the classroom to adapt its behaviors and engage in dynamic interactions. ClassMeta supports classroom participation through various interactions, including asking questions, raising questions, reminding students of key points, and facilitating discussions. The agent's behaviors are designed to encourage active participation and create a positive behavioral norm in the classroom. The design of ClassMeta leverages the capabilities of large language models (LLMs) to generate contextually coherent responses and simulate diverse human behaviors. The agent's role is configured to resemble that of a typical student with limited knowledge, rather than an all-knowing expert, to ensure its behaviors are realistic and conducive to classroom dynamics. A comparative user study was conducted to evaluate the effectiveness of ClassMeta in improving classroom participation. The study found that students in the ClassMeta condition showed increased attention to the agent, better note-taking quality, and more active participation in discussions compared to the baseline condition. The results suggest that ClassMeta can effectively promote classroom participation and enhance the learning experience in VR classrooms.