The study of both empirical and intelligent modeling is vital for enhancing the prediction, optimization, and control of the hydrothermal liquefaction (HTL) processes, which will ultimately lead to more efficient and sustainable biocrude production from HTL of biomass. This Chapter provides a comprehensive overview of the latest advancements in the predictive modeling for HTL product yield. Four primary empirical models are explored, namely kinetic, additive, response surface, and mixture design models. Kinetic and additive models are usually only employed to predict HTL product yields, while response surface models optimize specific HTL conditions and mixture design models optimize feedstock’s biochemical compositions to maximize the product yield. The intelligent modeling employs machine learning algorithms such as Random Forest and Gradient Boosting Regression for predicting biocrude yield and quality, demonstrating superiority over empirical equations. The coupling of Particle Swarm Optimization algorithm with prediction algorithms has facilitated a smart and adequate identification of optimal feedstock characteristics and HTL conditions. The emergence of open-source software tools has markedly reduced laboratory trial times, expediting HTL-based biofuel production. Future research should focus on refining machine learning algorithms for increasing accuracy, exploring other machine learning-based optimization methods (beyond Particle Swarm Optimization), and combining intelligent models with real-time product quality monitoring techniques such as multivariate regression of spectroscopy data. Such integration is crucial for an intelligent control of continuous HTL processes, a critical step towards HTL technique commercialization.