Geopolymer concrete (GC) is successful low carbon footprint alternative material to traditional ordinary Portland cement concrete (OPC) due to its excellent mechanical properties and reduced carbon emissions. Recent progress has been made in predicting the strength of GC using machine learning and deep learning techniques. Further investigation is required to comprehend the impact of particle packing density on GC models. This study develops machine learning and deep learning models to envisage the compressive strength (CS) of GC, incorporating packing density as an input feature. Specifically, three machine learning techniques such as multiple linear regression (MLR), artificial neural network (ANN), and support vector regressor (SVR), alongside a deep learning approach using long short-term memory networks (LSTM), are employed to estimate GC CS. The accuracy of each method is evaluated in a comparative study, revealing that LSTM offers the moast precise predictions, achieving R2 value of 0.942 and a mean square error (MSE) of 9.083. In comparison, the machine learning models demonstrated lower accuracy in estimating CS.