Accurate prediction of reaction performance is pivotal in designing efficient chemical synthesis pathways, particularly in drug discovery and chemical research. Machine learning approaches have emerged as powerful tools for predicting key reaction parameters, including yield, activity, selectivity, activation energy, and transition states. Predicting reaction yield, a measure of the efficiency of converting reactants into the desired products, is especially important for reducing experimental costs and time. Advanced models employing quantification techniques, molecular fingerprints, graph neural networks, and Transformer architectures have shown a great promise in predicting reaction outcomes. In addition to yield prediction, machine learning methods are increasingly applied to optimize reaction activity, selectivity, and energy profiles, enhancing the understanding and control of chemical processes. Traditional trial-and-error methods for optimizing reaction conditions are being replaced by artificial intelligence (AI)-driven approaches, which streamline experimental workflows and improve efficiency. These advancements highlight the transformative potential of AI in reaction performance prediction and condition optimization, paving the way for innovation in synthetic chemistry and related fields.