Today, Deep Learning technology has garnered significant attention from educational researchers, due to its ability to learn from enormous volumes of data and make accurate predictions. Predicting student academic performance is important in educational systems, enabling early interventions and personalized support. This paper presents an experimental study comparing deep learning models to predict academic performance. Experiments carried out include training and testing of models such as Multi-layer Perceptron, Recurrent Neural Networks, and Convolutional Neural Networks. In addition, the Logistic Regression model is used as a baseline for comparison with deep learning-based models. The efficacy of these models in predicting student performance is evaluated using a comprehensive dataset that includes demographic information of students and historical academic records. Experimental results indicate that the Multi-layer Perceptron model provides superior predictive performance, compared to other models in the majority of the metrics tested. A comparison of similar studies on academic performance prediction using Multi-layer perception in terms of accuracy is also presented. This study highlights the potential of deep learning in predicting university students’ academic performance.