Modern CDMA systems are required to provide differentiated services to support heterogeneous traffic while existing rate allocation schemes are designed for homogeneous users. We propose different utility functions to adapt to distinct user QoS requirements. Our goal is to maximize the system-wide aggregated utilities. A suboptimal solution of this NP-hard nonlinear programming problem is achieved by implementing GA-SA hybrid algorithm. Simulation results verify that our scheme allocates rates in such a way that all types of services are satisfied in balance while a high throughput is still maintained therefore more types of services are accommodated in the system.