City courier service still meets inefficiency, especially while handling multi destination orders. In the existing system, single transaction order means delivering packet from one sender to one destination. So, when a customer needs to send more than one packets to more than one destination, customer must create order one by one. Then a driver can execute only one delivery order. Besides that, a driver also cannot deliver packets that their destination is close to each other because these packets are ordered by different customers. Based on this problem, in this work, we develop dispatch model that supports combined shipping model. In combined shipping model, customer can create one order that contains more than one destination. Meanwhile, a driver can execute more than one packet concurrently. This work is a continuation of our previous works in combined shipping service and scheduled shipping service. In this work, we develop the dispatch model by combing and modifying k-medoid method and least cost method. In our previous work, the world is a square shaped city. In this work, the virtual work is a circle shaped city. This work is also the improvement of our previous work that uses sequential least cost and Round Robin method. This model then is implemented into city courier simulation application. Based on the simulation result, this proposed model performs better than the conventional method in reducing the total driver distance. By using these proposed models, the total travel distance reduction is above 60 percents. The reduction in total driver distance has positive relation with reduction in total cost. These proposed models also reduce travel distance disparity among drivers as it becomes problems in the previous work. The travel distance standard deviation of least first-medoid model is below 5 kilometer while travel distance standard deviation of Round Robin-medoid model is below 10 percent.