Climate change impacts forest functioning and dynamics, and large uncertainties remain regarding the interactions between species composition, demographic processes and environmental drivers. There are few robust tools available to link these processes, which precludes accurate projections and recommendations for long-term forest management. Forest gap models present a balance between complexity and generality and are widely used in predictive forest ecology. However, their relevance to tackle questions about the links between species composition, climate and forest functioning is unclear. In this regard, demonstrating the ability of gap models to predict the growth of forest stands at the annual parameterization scale resolution—representing a sensitive and integrated signal of tree functioning and mortality risk—appears as a fundamental step. In this study, we aimed at assessing the ability of a gap model to accurately predict forest growth in the short term and potential community composition in the long term, across a wide range of species and environmental conditions. To do so, we present the gap model ForCEEPS, calibrated using an original parameterization procedure for the main tree species in France. ForCEEPS was shown to satisfactorily predict forest annual growth (averaged over a few years) at the plot level from mountain to Mediterranean climates, regardless of the species. Such an accuracy was not gained at the cost of losing precision for long-term predictions, as the model showed a strong ability to predict potential community compositions. The mechanistic relevance of ForCEEPS parameterization was explored by showing the congruence between the values of key model parameter and species functional traits. We further showed that accounting for the spatial configuration of crowns within forest stands, the effects of climatic constraints and the variability of shade tolerances in the species community are all crucial to better predict short-term productivi