An improved firefly algorithm is investigated with the aid of genetic algorithm (GA). The proposed algorithm improves the load ability of power system with unified power flow controller (UPFC). Random movement factor of firefly algorithm is improved by hybridizing the GA with classical firefly algorithm. In firefly algorithm, the next movement of firefly is depends on the movement factor which is determined by randomly so the best movement of firefly is possibility to fails by the distribution of random number. Thus, the best location of and capacity of UPFC can never able to recognize accurately. So in this paper, a GA based optimization algorithm is used to determine the optimal random movement factor of fireflies. Thus, the optimal location and capacity of UPFC is determined efficiently when compared to traditional fire fly algorithm. The proposed method implemented in MATALB and the optimal location and capacity of UPFC is examined as per the variation of voltage, power loss and power balance of the network. The load power control performance of proposed method is compared with classical firefly algorithm.