Industry 4.0 enables data-driven, automated optimization of thermochemical processes. We investigate machine-learning models to predict syngas composition in biomass co-gasification, supporting real-time control and waste valorization. Using experimental datasets, we trained and evaluated Artificial Neural Networks (ANN), Gaussian Process Regression (GPR), and Gradient Boosting Regression (GBR). Model performance was assessed by the coefficient of determination (R²) and error metrics. GBR delivered the most accurate predictions (R²> 0.85), followed by GPR (≈0.80) and ANN (≈0.79). These results indicate that gradient-boosted ensembles are robust for capturing nonlinear relationships among feedstock properties and operating conditions. Embedding such models into Industry 4.0 architectures can continuously update predictions from live data, enabling tighter process control and more consistent syngas quality. This study demonstrates the practical value of machine learning for sustainable energy and biomass waste management.