Our investigation aimed to propose a new single-document extractive text summarization model, which consists of a classifier and a summary component based on pre-trained clustering models. First, we train the classifier on the training data set, then we train the clustering models on the subtraining data sets, and split from the entire training data set. In the summary process, the summary model uses the classifier to predict the input text label and then uses this label to choose the corresponding clustering model for summarizing. The model’s numerical test results on the Vietnamese data set based on ROUGE-1, ROUGE-2, and ROUGE-L are 51.50%, 16.26%, and 29.25%, respectively. In addition, our model can perform well on cost-effective resources like an ARM CPU to summarize large amounts of documents.