This study aims to develop a new DiffServ queue model and AQM (Active Queue Management) model to improve the quality of service in real-time internet applications. This model, called MCPRQ (Multi-Criteria Priority RED Queuing), aims to provide more effective queue management by evaluating packets according to their priority levels, sizes, and waiting times within the scope of the DiffServ architecture. This evaluation is performed using the Analytical Hierarchy Process (AHP) and integrated with the RED (Random Early Detection) algorithm to provide a solution to increase the quality of service in queue management. The MCPRQ model has been tested with the OMNeT++ simulator in IPv6 networks and has achieved successful results compared to commonly used queue structures. Its low packet loss has attracted attention, especially in low-density networks, and low average delay in high-density networks. This shows that MCPRQ offers a significant advantage in flexibility and scalability. As a result, the MCPRQ model effectively manages congestion in medium and high-density networks while providing better performance by preserving the quality of service in real-time applications.