Text summarization is a crucial task in natural language processing (NLP), where large language models (LLMs) like Llama 2, T5, and BART have shown significant promise. This study focuses on fine-tuning these models for text summarization using the CNN/Daily Mail dataset. By leveraging transformer-based architecture, the authors aim to generate concise and accurate summaries of news articles. The study compares the performance of these models using the ROUGE metric, assessing both their advantages and limitations. The results demonstrate that BART outperforms Llama 2 and T5 in producing coherent summaries, particularly for short news articles. This journal also delves into the challenges of fine-tuning large models, such as managing overfitting and computational resource constraints. The significance of hyperparameter streamlining is underscored, as it assumes a basic part in upgrading model execution. The authors likewise investigate different systems, including move learning and information increase, to work on rundown quality. Additionally, the study identifies future directions for research, such as incorporating user feedback and developing hybrid models that combine extractive and abstractive summarization techniques. Overall, the findings contribute to the ongoing discourse in the field, offering insights that may enhance the efficacy of text summarization systems.