This study investigates the influence of ZycoTherm, a warm mix asphalt additive, on the physical, rheological, and mechanical properties of 60/70 penetration grade bitumen, complemented by machine learning (ML)-based predictive modeling. Six modified bitumen samples were prepared with ZycoTherm dosages of 0%, 0.05%, 0.08%, 0.10%, 0.15%, and 0.20% by weight. Standardized laboratory tests were conducted to evaluate penetration, softening point, viscosity, stripping resistance, complex shear modulus (G*), and Marshall stability. The penetration results indicated an increase up to 0.08% ZycoTherm, enhancing flexibility, followed by a decrease, suggesting higher rigidity at greater dosages. The softening point and viscosity improved at moderate dosages but showed slight deterioration beyond the optimum. Stripping resistance significantly improved with higher ZycoTherm dosages (0.15–0.20%), while Dynamic Shear Rheometer results confirmed increased stiffness and reduced rutting susceptibility. Marshall stability reached optimal performance at 0.15% dosage, while flow values remained within safe ranges, indicating balanced performance. To advance the analysis, a dataset comprising 100 literature-derived samples was compiled using key parameters such as ZycoTherm dosage, penetration, softening point, viscosity, stripping value, flow, and shear modulus to predict Marshall stability. Multiple ML algorithms including Artificial Neural Networks (ANN), Support Vector Regression (SVR), and Gaussian Process Regression (GPR) were implemented. Among these, Bayesian GPR provided the most robust predictive capability with an R² value exceeding 0.92, effectively capturing the nonlinear interactions between input parameters and stability response. The ML models were further validated against independent studies, confirming their reliability in extrapolating performance trends. The findings highlight ZycoTherm as a promising additive for enhancing durability, workability, and moisture resistance of asphalt pavements, while ML frameworks provide a scalable tool for optimizing additive dosage and forecasting long-term performance.