<p>This study developed a machine learning–based framework to estimate the factor of safety (FOS) in slope stability analysis, providing a computationally efficient alternative to traditional limit equilibrium methods. Five models, decision tree, adaptive boosting, categorical boosting, artificial neural network, and long short-term memory, were trained and evaluated. Two interpretability methods, SHapley Additive exPlanations and local interpretable model-agnostic explanations, were incorporated to improve transparency. A large-scale dataset of 19,128 samples was used, including key geotechnical and geometric parameters such as slope angle, friction angle, cohesion, slope height, unit weight, horizontal seismic acceleration, and the seismic acceleration ratio. Among the models, categorical boosting, artificial neural network, and long short-term memory achieved the best performance, with <i>R</i>² values of 0.999 on training data and up to 0.999 on testing data, along with minimal root mean square error and mean absolute error. The interpretability analysis showed that the stability number (<i>c</i>/γH) had the greatest influence on FOS, followed by friction and slope angles. Seismic loading was found to reduce stability, aligning with geotechnical principles. The integration of predictive and interpretability methods enhanced model reliability, offering a practical, accurate, and explainable approach for slope stability assessment.</p>