Multilevel Inverters play key role in grid integration of photovoltaic (PV) systems. Although power semiconductor devices in multilevel inverters represent the most expensive and vulnerable parts according to the recent reliability surveys. Thermal stresses in power semiconductor devices represent the main cause of their failures. Therefore, improved controllers with lower thermal stresses are highly demanded. Conventional finite control set model predictive control (MPC) utilizes a high sampling rate to select the optimum switching state from all possible states. Consequently, high power losses and high thermal stresses are produced in this type. This paper introduces an improved model predictive controller (MPC) for preserving high reliability for grid connected single phase full bridge five level T-type inverter. The proposed controller selects the optimum switching sequence to drive the inverter from a set of switching sequences with an optimized switching transitions. The proposed MPC features constant switching frequency with minimized switching transitions of power devices and voltage balance of DC-link capacitors. The results of the designed and implemented case study are presented to validate the superiority of the proposed controller.