Voltage collapse tends to occur due to the voltage instability created during large faults. As a last resort, under-voltage load shedding (UVLS) is performed after all the available power operation and control mechanisms have been exhausted, in order to save a system from voltage collapse. Load shedding techniques have advanced from the conventional and adaptive techniques that were less optimal compared to computational intelligence techniques (CITs). Recent works have identified hybrid algorithms to give more optimal solutions for UVLS problems with multi-objective functions. In this paper, a novel hybrid ABC-PSO algorithm, adapted from a software estimation project, was used to perform UVLS on a modified IEEE 14-bus system. The modified system was made to reflect a real-life system with considerations on different load models. From the FVSI ranking used, 8 overload conditions were imposed on the system ranging from 105% to 140% loading. The load shedding was performed in accordance to decentralized relay settings of 3.5 seconds, 5 seconds and 8 seconds, which resulted in an overall 99.32% recovery of voltage profiles. The proposed hybrid ABC-PSO algorithm was able to shed optimal amounts of load, giving an 85.50% post-contingency load, compared to GA’s 77.04%, ABC-ANN at 84.03% and PSO-ANN at 80.96%. This study was simulated on MATLAB software, using the Power System Analysis Toolbox (PSAT) graphical user and command line interfaces.