We are interested in the prediction of option prices, which are financial derivatives offering a different risk-return profile compared to stocks—typically involving lower risk but also more limited gains. The classical Black-Scholes model has long been used for option pricing, but it relies on strong assumptions such as constant volatility, which often do not hold in real markets. In this project, we develop alternative models that relax these assumptions and better capture market behavior for FAANG companies. Using real market data, we explore regression-based methods, as well as ensemble models Random Forest, XGBoost and neural network, to improve prediction accuracy and better align with observed option prices. Dataset is available on Yahoo Finance.