Advanced computational fluid dynamics (CFD) models of gas release and dispersion (GRAD) have been developed, tested, validated and applied to the modeling of various industrial real-life indoor and outdoor flammable gas (hydrogen, methane, etc.) release scenarios with complex geometries. The user-friendly GRAD CFD modeling tool has been designed as a customized module based on the commercial general-purpose CFD software, PHOENICS. Advanced CFD models available include the following: the dynamic boundary conditions, describing the transient gas release from a pressurized vessel, the calibrated outlet boundary conditions, the advanced turbulence models, the real gas law properties applied at high-pressure releases, the special output features and the adaptive grid refinement tools. One of the advanced turbulent models is the multifluid model (MFM) of turbulence, which enables to predict the stochastic properties of flammable gas clouds. The predictions of transient threedimensional (3D) distributions of flammable gas concentrations have been validated using the comparisons with available experimental data. The validation matrix contains the enclosed and nonenclosed geometries, the subsonic and sonic release flow rates and the releases of various gases, e.g., hydrogen, helium, etc. GRAD CFD software is recommended for safety and environmental protection analyses. For example, it was applied to the hydrogen safety assessments including the analyses of hydrogen releases from pressure relief devices and the determination of clearance distances for venting of hydrogen storages. In particular, the dynamic behaviors of flammable gas clouds (with the gas concentrations between the lower flammability level (LFL) and the upper flammability level (UFL)) can be accurately predicted with the GRAD CFD modeling tool. Some examples of hydrogen cloud predictions are presented in the paper. CFD modeling of flammable gas clouds could be considered as a costeffective and reliable tool for environmental assessments and design optimizations of combustion devices. The paper details the model features and provides currently available testing, validation and application cases relevant to the predictions of flammable gas dispersion scenarios. The significance of the results is discussed together with further steps required to extend and improve the models.