Abstract In Latin America, distribution networks are mostly radial, with high energy losses, representing about 16% of the region's electricity consumption. In Cuba, one of the alternatives that has been proposed to reduce these losses is the distribution of photovoltaic (PV) power plants of up to 3 MWp in them. This paper presents a new hybrid optimization algorithm to localize and size PV power plants to minimize energy losses in radial distribution networks. The method localizes the power plants using Empirical Discrete Metaheuristic in conjunction with a novel strategy that reduces the execution time of the algorithm. The sizes are found using the Coordinate Search Method, which performs one‐dimensional searches in the directions of the variables of the objective function. The algorithm was run on a 33‐bus IEEE test system in which up to five power plants were optimally installed, taking into account load variations at the distribution level and changes in the weather conditions under which the power plants operate. It should be noted that the execution times for each scenario did not exceed 30 s. The algorithm was validated by comparing it with other methods found in the literature and proved to be more efficient than them.