The global shift towards renewable energy sources has deepened interest in biomass as a viable alternative to fossil fuels. Biomass is the only renewable carbon source. Biomass gasification in fluidized bed reactors is a promising route to produce syn-gas. Despite its potential, challenges persist in effectively converting biomass into usable fuels, particularly due to tar production in fluidized bed reactors. Understanding the coupling between gas-phase reactions and multiphase hydrodynamics is essential for optimization of fluidized bed reactors. Experiments alone are insufficient for offering detailed physical insights, due to the opaque nature of the reactor and complex interactions necessitate computational approaches. But, traditional Computational Fluid Dynamics (CFD) simulations are computationally expensive, prompting the exploration of more efficient alternatives. Especially, the computational time for calculating the species reaction rate in CFD simulations makes it expensive. In this study, we propose a novel approach utilizing the Machine Learning (ML) model to represent the gas-phase reactions in the fluidized bed reactor offering faster and more cost-effective solutions compared to CFD simulations. Specifically, we propose a Machine Learning (ML) model to predict species compositions along the reactor length. This ML-based model is assessed by comparing its predictions with the CFD simulations using detailed kinetics. In addition, we propose to develop a ML model to predict the species reaction rate and integrate this model in CFD simulations. This ML-based approach significantly accelerates predictions of biomass thermochemical conversion while maintaining accuracy comparable to detailed kinetics-based CFD simulations.