The present disclosure provides a method for multi-group audio-video interaction, including: receiving a virtual group creation request initiated by a first user, creating a first virtual group, and determining a first session group the first user belonging to; associating the first session group with the first virtual group, and adding all session users in the first session group to the first virtual group; determining a second session group a second user belonging to, when receiving a virtual group join request initiated by the second user; associating the second session group with the first virtual group, and adding all session users in the second session group to the first virtual group; and sending a user list of the second session group to other session users in the first virtual group, to enable the other session users to incrementally pull and play an audio-video stream according to the user list.