Rationalizing the influence of the solvent on electrochemical reaction energetics is a central challenge in our understanding of electrochemical interfaces. To date, it is unclear how well existing methods predict solvation energies at solid/liquid interfaces since they cannot be assessed experimentally. Ab initio molecular dynamics (AIMD) simulations present a physically highly accurate, but also a very costly approach. In this work, we employ extensive AIMD simulations to benchmark solvation at charge-neutral metal/water interfaces against commonly applied continuum solvent models. We consider a variety of adsorbates including *CO, *CHO, *COH, *OCCHO, and *OH on Cu, Au, and Pt facets solvated by water. The surfaces and adsorbates considered are relevant, among other reactions, to electrochemical CO2 reduction and the oxygen redox reactions. We determine directional hydrogen bonds and steric water competition to be critical for a correct description of solvation at the metal/water interfaces. As a consequence, we find that the most frequently applied continuum sol- vation methods, which do not yet capture these properties, do not presently provide more accurate energetics over simulations in vacuum. We find most of the computed benchmark solvation energies to linearly scale with hydrogen bonding or competitive water adsorption, which strongly differs across surfaces. Thus, we determine solvation energies of adsorbates to be non-transferable between metal surfaces in contrast to standard practice.