Machine translation (MT), which refers to automated systems that generate translations between natural languages, either with or without human intervention, has experienced remarkable advancements over the years. Translating Marathi to English is crucial for bridging communication gaps between regional and global audiences, enabling access to diverse information, and fostering cross-cultural exchange. The major development in improving the credibility of machine translation was achieved through the use of Neural Machine Translation (NMT). This NMT is based on a combination of recurrent or transformer-based neural networks to produce more contextual and natural translations. While this study primarily focuses on the language translation from Marathi to English, the model designs can be applied to other language pairs with similar structures. The study discusses the different methodologies of NMT used by numerous researchers in this field and highlights their respective advantages and drawbacks. Three models, the Long-Short Term Memory (LSTM) model, the Transformer architecture-based model, and Google’s multilingual T5 (mT5) model are applied in this study, and their performances are compared and discussed.