The takeout platform continues to improve the quality of customer services and reduce costs while ignoring the interests of riders. Considering the dilemma of riders, the incentive mechanism of reward and punishment is introduced, and the takeout order allocation model is established with the platform cost, customer satisfaction and rider incentive income as the objective. Based on the order allocation mode of combination of dispatching and grabbing, considering the convenience, distance and capability, the utility function is constructed to describe the preference and matching degree when grabbing order autonomously, and the Logit discrete selection model is used to simulate order grabbing behavior. The adaptive variable neighborhood search algorithm is designed to implement the platform dispatch allocation and path optimization. The experimental results show that the delivery scheme based on the order allocation mode of order-dispatching and grabbing effectively reduces the overall distribution cost, improve customer satisfaction and rider revenue compared with the mode of order-dispatching. The research results extend the theoretical research on the takeout order allocation mode, and provide reference for the benign development of takeout industry.