Abstract: Tekes County, located in northern Tianshan Mountains, is dominanted by low mountainous and hilly terrain, where regional landslide disasters are highly developed. In this study, 10 factors impacting the development of landslide disasters, including elevation, slope gradient, slope aspect, terrain curvature, relative relief, engineering geological rock group, fault, NDVI, river system and road, are selected to construct the susceptibility evaluation indexes of landslides in Tekes County. First, the information model is applied for landslide susceptibility assessment. Subsequently, based on the susceptibility zoning results from the information model, non-landslide sample data are randomly selected in a proportion equal to the number of landslides, which is then combined with the landslide samples. Finally, the ensemble learning algorithm, specifically the gradient boosting decision tree (GBDT) and extreme gradient boosting(XGBoost) models are employed for landslide susceptibility assessment. Based on the susceptibility indexes, the study area is divided into five susceptible zones, i.e. extremely low, low, medium, high and extremely high. The accuracy of evaluation results is analyzed using ROC curves and landslide density. The results indicate that the XGBoost model combined with information method achieves the highest prediction accuracy, with the AUC value of 0.91, and produces the most reasonable susceptibility zoning result. There are 144 landslide disasters(86.23% of the total) located within the high and extremely high susceptible zones which are mainly concentrated in the low mountainous and hilly areas of northern Tekes County and areas with intensive human engineering activities.