Based on the 2019 assessment of the Global Carbon Project, the ocean took up on average, 2.5 ± 0.6 PgC yr−1 or 23 ± 5% of the total anthropogenic CO2 emissions over the decade 2009–2018. This sink estimate is based on simulation results from global ocean biogeochemical models (GOBMs) and is compared to data-products based on observations of surface ocean pCO2 (partial pressure of CO2) accounting for the outgassing of river-derived CO2. Here we evaluate the GOBM simulations by comparing the simulated surface ocean pCO2 to observations. Based on this comparison, the simulations are well-suited for quantifying the global ocean carbon sink on the time-scale of the annual mean and its multi-decadal trend (RMSE <20 μatm), as well as on the time-scale of multi-year variability (RMSE <10 μatm), despite the large model-data mismatch on the seasonal time-scale (RMSE of 20–80 μatm). Biases in GOBMs have a small effect on the global mean ocean sink (0.05 PgC yr−1), but need to be addressed to improve the regional budgets and model-data comparison. Accounting for non-mapped areas in the data-products reduces their spread as measured by the standard deviation by a third. There is growing evidence and consistency among methods with regard to the patterns of the multi-year variability of the ocean carbon sink, with a global stagnation in the 1990s and an extra-tropical strengthening in the 2000s. GOBMs and data-products point consistently to a shift from a tropical CO2 source to a CO2 sink in recent years. On average, the GOBMs reveal less variations in the sink than the data-based products. Despite the reasonable simulation of surface ocean pCO2 by the GOBMs, there are discrepancies between the resulting sink estimate from GOBMs and data-products. These discrepancies are within the uncertainty of the river flux adjustment, increase over time, and largely stem from the Southern Ocean. Progress in our understanding of the global ocean carbon sink necessitates significant advancement in modeling and observing the Southern Ocean carbon sink including (i) a game-changing increase in high-quality pCO2 observations, and (ii) a critical re-evaluation of the regional river flux adjustment.