The overwhelming volume of information today makes it difficult for readers to effectively process key content. Manual text summarization is time-consuming and impractical for large volumes of text. Automatic Text Summarization addresses these challenges by using NLP techniques to generate concise summaries, simplifying lengthy content for quicker consumption. This paper presents a three-phase Hybrid Model for Text Summarization (HMTS). In the first phase, the original documents are pre-processed to ensure compatibility with the subsequent phases. In the next phase, an extractive approach is employed, wherein various classification models are tested on the pre-processed data to classify sentences based on their importance. The significant sentences identified during the extractive phase are then input into the BART transformer model in the third phase to produce the final candidate summary. The feasibility of the proposed model is evaluated using the “BBC News dataset,” and accuracy scores are computed using the ROUGE evaluation metric. The experimental results reveal the better performance of the proposed model over existing models. Experimental results on the BBC News dataset indicate that CNN+BART surpasses traditional approaches, achieving improvements of 2.07% in ROUGE-1, 1.20% in ROUGE-2, and 7.89% in ROUGE-L scores. Additionally, the model demonstrates higher precision, recall, and F1-score, ensuring more accurate and meaningful summaries. This approach enhances text summarization by improving the relevance, coherence, and overall quality of generated summaries.