Text summarization role in Natural Language Processing (NLP) is considered as highly important and crucial. This process uses large volumes of text data as input. This text data is condensed into brief summaries by preserving essential information. Telugu language is one of the major Dravidian languages spoken by over 80 million people across the world. This language remains as low-resource language where the available dataset is limited. In this paper the challenge of summarizing Telugu text using is addressed by using state-of-the-art deep learning techniques. This work explores both extractive summarization and abstractive summarization approaches. Further fine-tuning the transformer-based models such as mBART and mT5 on collected Telugu text data collected. This proposed model exhibits satisfactory performance when compared to other traditional techniques and existing techniques. Further this model has shown significant improvements in ROUGE scores. This work also presents a qualitative evaluation of generated summaries, by highlighting the fluency and consistency of the outputs. The study enables to learn advancing NLP for languages that are underrepresented and becomes as a source for upcoming research in multilingual text summarization.