The cycling of carbon in the oceans is affected by feedbacks driven by changes in climate and atmospheric CO 2 . Understanding these feedbacks is therefore an important prerequisite for projecting future climate. Marine biogeochemistry models are a useful tool but, as with any model, are a simplification and need to be continually improved. In this study, we coupled the Finite-volumE Sea ice-Ocean Model (FESOM2.1) to the Regulated Ecosystem Model version 3 (REcoM3). FESOM2.1 is an update of the Finite-Element Sea ice-Ocean Model (FESOM1.4) and operates on unstructured meshes. Unlike standard structured-mesh ocean models, the mesh flexibility allows for a realistic representation of small-scale dynamics in key regions at an affordable computational cost. Compared to the previous coupled model version of FESOM1.4-REcoM2, the model FESOM2.1-REcoM3 utilizes a new dynamical core, based on a finite-volume discretization instead of finite elements, and retains central parts of the biogeochemistry model. As a new feature, carbonate chemistry, including water vapour correction, is computed by mocsy 2.0. Moreover, REcoM3 has an extended food web that includes macrozooplankton and fast-sinking detritus. Dissolved oxygen is also added as a new tracer. In this study, we assess the ocean and biogeochemical state simulated with FESOM2.1-REcoM3 in a global set-up at relatively low spatial resolution forced with JRA55-do atmospheric reanalysis. The focus is on the recent period (1958-2021) to assess how well the model can be used for present-day and future climate change scenarios on decadal to centennial timescales. A bias in the global ocean-atmosphere preindustrial CO 2 flux present in the previous model version (FESOM1.4-REcoM2) could be significantly reduced. In addition, the computational efficiency is 2-3 times higher than that of FESOM1.4-REcoM2. Overall, it is found that FESOM2.1-REcoM3 is a skilful tool for ocean biogeochemical modelling applications.