Current swarm intelligence evacuation models only consider single classic swarm intelligence, which is insufficient to describe the complex behavior characteristics of crowd evacuation. In addition, these models rarely take into consideration the impact of crowd chaos on crowd evacuation. In order to study the swarm intelligence evacuation model describing the evacuation behaviors of different groups, by integrating various swarm intelligence algorithms, and taking into account the impact of crowd chaos on evacuation, a crowd evacuation model based on hybrid artificial bee colony-bat algorithm with entropy correction was proposed. Firstly, the density-based spatial clustering of applications with noise (DBSCAN) algorithm was used for group partition. The evacuees were divided into group leader, group members, and disorganized people. Next, according to the characteristics of each type of evacuees, the group leader was described based on bat algorithm, the group members based on artificial bee colony algorithm, and the disorganized people based on particle swarm optimization (PSO). Finally, the evacuation entropy that quantitatively describes the degree of crowd chaos was introduced to correct the position of the group leader, and the evacuation model based on hybrid artificial bee colony-bat algorithm with entropy correction was thus constructed. Simulation results show that the model could well simulate group evacuation, which was basically consistent with the real shape of group evacuation, and the evacuation efficiency was improved by means of group evacuation to some extent. With the introduction of the evacuation entropy for correction, the group leader could guide the group members to avoid the chaotic area ahead, prevent the excessive concentration of evacuees, and enhance the safety and rapidity of evacuation.