Providing good QoS in Cellular IP networks is one of the important issues in order to improve the performance of the Cellular IP network. Resource reservation is one of the methods used in achieving this goal and is proven to be effective. Resources of Cellular IP network that can be reserved are bandwidth, buffer and CPU cycles. Router CPU cycle is the time taken to process a packet before forwarding it to the next router (hop). This paper is proposing a model for CPU cycle optimization of routers for real-time flows in Cellular IP network. The model applies Genetic Algorithm as a soft computing tool to optimize the CPU cycles and reduces the flow processing time at each router in the route taken by a flow. Simulation experiments show the efficacy of the model.