Vegetation regulates land-atmosphere, water, and energy exchanges and is an essential component of land-surface models (LSMs). However, LSMs have been handicapped by assumptions that equate acclimated photosynthetic responses to the environment with the fast responses observable in the laboratory. The effects of acclimation can be taken into account by including PFT-specific values of photosynthetic parameters, but at the cost of increasing parameter requirements. Here, we develop an alternative approach for including acclimation in LSMs by adopting the P model, an existing light-use efficiency model for gross primary production (GPP) that implicitly predicts the acclimation of photosynthetic parameters on a weekly to monthly timescale via optimality principles. We demonstrate that it is possible to explicitly separate the fast and slow photosynthetic responses to environmental conditions, allowing the simulation of GPP at the sub-daily timesteps required for coupling in an LSM. The resulting model reproduces the diurnal cycles of GPP recorded by eddy-covariance flux towers in a temperate grassland and boreal, temperate and tropical forests. The best performance is achieved when biochemical capacities are adjusted to match recent midday conditions. Comparison between this model and the operational LSM in the European Centre for Medium-range Weather Forecasts climate model shows that the new model has better predictive power in most of the sites and years analyzed, particularly in summer and autumn. Our analyses suggest a simple and parameter-sparse method to include both instantaneous and acclimated responses within an LSM framework, with potential applications in weather, climate, and carbon-cycle modeling.