The development of an accurate design-oriented model for fiber-reinforced geopolymer concrete (FRGPC) is crucial for ensuring the reliable design of this composite system. The empirical approaches have limitations in their ability to consider every influencing aspect simultaneously when forecasting the strength properties of FRGPC. This study introduces a novel hybrid model for predicting the strength characteristics of FRGPC based on ground granulated blast furnace slag (GGBS) by integrating extreme gradient boosting (XGBoost) and particle swarm optimization (PSO). XGBoost is a machine learning (ML) technique that employs a non-linear mapping function to generalize and derive strength property results from input data. A swarm-based metaheuristic is also utilized to optimize the ML algorithm by fine-tuning its hyperparameters. The proposed hybridization of XGBoost and PSO, designed to forecast the strength characteristics of FRGPC encompassing properties such as compressive strength, split tensile strength, and flexural strength, has been developed and validated using a dataset obtained from the literature. The effectiveness of the proposed hybrid model was assessed using various statistical measures, including R2, RMSE, MAE, MSE, and MAPE. The statistical analysis results demonstrated that the proposed hybrid model provided more accurate predictions for the strength properties of GGBS-based FRGPC compared to the XGBoost model. The assessment outcomes also indicated that the hybrid model demonstrated strong agreement between the actual and predicted strength values, characterized by a high correlation coefficient and lower error values. Hence, the developed hybrid model contributes to the advancement of predictive modeling in sustainable construction materials, providing a valuable tool for engineers and researchers involved in designing and optimizing FRGPC structures.