In NLP, identifying sarcasm in text poses a significant challenge. Accurate determination between sarcastic or non-sarcastic statements is vital for precise results in sentiment analysis and opinion mining, among many other NLP applications. This study compares several models designed to effectively identify sarcastic versus non-sarcastic content to analyze this complex issue. We evaluate traditional approaches like logistic regression, Support Vector Machine, and deep learning models like CNNs and RNNs. Furthermore, utilizing a sizable dataset of annotated sarcastic versus non-sarcastic speeches trains deep learning methods, including transformers such as BERT and GPT3. Our research evaluates the efficacy of various feature extraction and pre-processing approaches on our model's performance. The paper deduces their strengths and weaknesses by analyzing different models' precision, recall, accuracy, and F1-score metrics.