This work delves into the application of transformer-based models to conduct a multifaceted analysis of challenges from the Codeforces platform. By harnessing the power of state-of-the-art transformer architectures such as BERT and Mistral-7B-Instruct-v0.2, the study aims to provide insights into challenge summarization, explanation, tag prediction, and rating prediction. The proposed work demonstrates significant advancements in prediction accuracy, achieving 98% accuracy for rating prediction and 73.85% for tag prediction, surpassing traditional models like GRU and LSTM. Through rigorous experimentation and evaluation, the findings elucidate the effectiveness of transformer models in addressing the unique linguistic and computational challenges inherent in analyzing code-related content. The findings offer valuable implications for enhancing the understanding of code-centric data and improving the efficiency of code-related tasks.