Among the most impactful aspects of future climate change is the enhancement of extreme precipitation, which will be critical for water resources and flood risk planning around the globe. Current projections of future enhancements of extreme precipitation come from general circulation models (GCMs), with the CMIP6 ensemble comprising the latest generation of global projections. However, the reliability of the GCMs to simulate the extreme events of interest, particularly those of a convective nature, is questionable. To address this issue, we evaluate extreme precipitation in CMIP6 models in the recent climate. This is achieved by comparison to various observational precipitation estimates, both global (TRMM, GPCP, CPC) and regional (PRISM). We compare the models and observations across the tail of the probability distribution of precipitation, for both daily and sub-daily extremes. The magnitude of daily precipitation extremes is of comparable magnitude between the models and observations. However, with decreasing interval over which precipitation is totaled, there is a larger underestimate of GCM precipitation: by a factor of about two for six-hourly extremes and by a factor of about four for three-hourly extremes. These comparisons indicate that the intensity of extreme precipitation is far too low in GCMs, particularly in the tropics. We subsequently analyze vertical velocity, conditioned onto extreme precipitation in various models. A strong correlation emerges between vertical velocity and the resolved component of precipitation (which constitutes the majority of total precipitation) during extreme events, illustrating that insufficient magnitudes of vertical velocity during extreme events are a large contributor to the underestimates of precipitation. The simulation of updrafts in GCMs in the present climate is related to the “dynamical” component of projected changes to extreme precipitation (the component arising from changes to vertical velocity), which previous studies have shown accounts for the model spread in projected changes to extreme precipitation. Improved representation in GCMs of updrafts during extreme events will lead to improved prediction of future changes to extreme precipitation.