Photovoltaic (PV) panels are commonly used as clean energy sources. However, their performance is sensitive to environmental conditions. In this work, a hybrid model of artificial neural networks (ANNs) and multiverse optimization (MVO)/genetic algorithm (GA) were implemented to predict PV output power, efficiency, and cell temperature. The main goal of the optimizer was to develop and update an ANN methodology based on training and forecasting. The numbers of neurons in the hidden layers, weights, and biases of the proposed ANNs were optimized with MVO and GA. The multilayer feedforward neural network (MFFNN) was used, and its accuracy was investigated through the results obtained from MFFNN-MVO and the MFFNN-GA models. Three parameters controlled the panel output power and temperature, and in turn, panel efficiency: ambient temperature, wind speed, and solar irradiance. The training and testing data were measured from a 4-kW PV plant installed in Shaqra City, Saudi Arabia, for two years. Moreover, the relationship between the PV panel efficiency and cell temperature was investigated. The PV efficiency relied on the sensitivity to the panel temperature factor (􀀀 0.06) and the maximum efficiency at 0 ⁰C (14.74). The efficiency of PV panels was predicted with a normalized root mean square error (NRMSE) of 3.65E-4 and 2.82E-4 for MFFNN-GA and MFFNN-MVO models, respectively.