One of the most important objectives of monetary institutions is to maintain price stability in countries or regions. Hyperinflation and deflation have adverse influences on economic development and potentially easily lead to a negative economic cycle or a sharp recession. Therefore, accurate inflation rate forecasting is essential in formulating a monetary policy. The Extreme Gradient Boosting (XGBoost) technique has recently become a popular and powerful tool in machine learning, which integrates weak learners and uses the regularization method to decrease overfitting. However, the use of XGBoost in forecasting inflations has not been broadly investigated. Thus, this study aims to utilize the XGBoost approach with the genetic algorithm to forecast inflations for three forecast horizons. The other six forecasting models—random forest (RF), least square support vector regression (LSSVR), backpropagation neural networks (BPNN), general regression neural networks (GRNN), deep belief networks (DBN), and long short-term memory (LSTM)—were employed to predict inflations with the same datasets. The numerical results indicate that the XGBoost models outperformed the other six forecasting models in all forecast horizons. Therefore, the XGBoost technique is a feasible and promising alternative for forecasting inflations.