High Mountain Asia (HMA) is among the most vulnerable water towers globally and yet future projections of water availability in and from its high‐mountain catchments remain uncertain, as their hydrologic response to ongoing environmental changes is complex. Mechanistic modeling approaches incorporating cryospheric, hydrological, and vegetation processes in high spatial, temporal, and physical detail have never been applied for high‐elevation catchments of HMA. We use a land surface model at high spatial and temporal resolution (100 m and hourly) to simulate the coupled dynamics of energy, water, and vegetation for the 350 km2 Langtang catchment (Nepal). We compare our model outputs for one hydrological year against a large set of observations to gain insight into the partitioning of the water balance at the subseasonal scale and across elevation bands. During the simulated hydrological year, we find that evapotranspiration is a key component of the total water balance, as it causes about the equivalent of 20% of all the available precipitation or 154% of the water production from glacier melt in the basin to return directly to the atmosphere. The depletion of the cryospheric water budget is dominated by snow melt, but at high elevations is primarily dictated by snow and ice sublimation. Snow sublimation is the dominant vapor flux (49%) at the catchment scale, accounting for the equivalent of 11% of snowfall, 17% of snowmelt, and 75% of ice melt, respectively. We conclude that simulations should consider sublimation and other evaporative fluxes explicitly, as otherwise water balance estimates can be ill‐quantified.