Diversity plays critical roles in ecosystem functioning, but it remains challenging to model phytoplankton diversity in order to better understand those roles and reproduce consistently observed diversity patterns in the ocean. In contrast to the typical approach of resolving distinct species or functional groups, we present a ContInuous TRAiT-basEd phytoplankton model (CITRATE) that focuses on macroscopic system properties such as total biomass, mean trait values, and trait variance. This phytoplankton component is embedded within a nitrogen–phytoplankton-zooplankton–detritus–iron model that itself is coupled with a simplified one-dimensional ocean model. Size is used as the master trait for phytoplankton. CITRATE also incorporates trait diffusion for sustaining diversity and simple representations of physiological acclimation, i.e., flexible chlorophyll-to-carbon and nitrogen-to-carbon ratios. We have implemented CITRATE at two contrasting stations in the North Pacific where several years of observational data are available. The model is driven by physical forcing including vertical eddy diffusivity imported from three-dimensional general ocean circulation models (GCMs). One common set of model parameters for the two stations is optimized using the Delayed-Rejection Adaptive Metropolis–Hasting Monte Carlo (DRAM) algorithm. The model faithfully reproduces most of the observed patterns and gives robust predictions on phytoplankton mean size and size diversity. CITRATE is suitable for applications in GCMs and constitutes a prototype upon which more sophisticated continuous trait-based models can be developed.