Text summarization produces a shortened or condensed version of input text highlighting its central ideas. Generating text summarization manually takes time and effort. This paper investigates several text summarization models based on neural networks, including extractive summarization, abstractive summarization, and abstractive summarization based on the re-writer approach and bottom-up approach. We perform experiments on the CTUNLPSum dataset in Vietnamese comprising 95,579 documents collected from commonly read Vietnamese online newspapers. The summarization model based on the bottom-up approach creates the best summaries. The F1-scores of ROUGE-1, ROUGE2, and ROUGE-L of the bottom-up approach are 0.598, 0.260, and 0.455, respectively.