This study compares the performance of five Large Language Models (LLMs)-ChatGPT, Google Gemini, Bing Copilot, Claude, and Cohere-in summarizing Indian legal text documents. We collect a diverse dataset of legal documents, preprocess them to remove noise, and tokenize them into sentences. Each LLM is then used to generate summaries, which are evaluated using standard metrics such as F1 Score, Recall, and Accuracy. Additionally, we solicit feedback from legal experts to assess the relevance, accuracy, and completeness of the summaries. Our results show that ChatGPT performed the best overall, with Google Gemini as a close second. These findings suggest that ChatGPT and Google Gemini are promising tools for summarizing Indian legal text documents. Further research is needed to explore the specific strengths and weaknesses of each LLM and address challenges such as domain-specific pretraining. This study contributes to the literature on LLMs in legal text summarization and provides guidance for future research and practice.
LLM, Natural Language Processing, Legal Text