Process-level uncertainties challenge system-level direct air capture assessments

System-level assessments of direct air capture (DAC) commonly rely on individual process models, overlooking the influence of model choice on system-level conclusions. Here, we quantify this impact by integrating weather-dependent performance data from seven solid-sorbent DAC models into an otherwise consistent energy system framework. DAC representation is the largest source of variation in levelized CO2 capture cost, accounting for 43% of the observed variation across the climate-renewable archetypes considered compared with 16% for climate and 6% for renewable-resource configuration. Model-dependent cost differences exceed 1000 € tCO2-1 at individual locations. Different model representations can also reverse siting preferences, substantially alter cost-optimal system design, and determine whether future CO2 capture-cost targets are achieved. Our results further demonstrate how individual process-level assumptions can propagate through the assessment, resulting in a major source of system-level uncertainty. The harmonized DAC performance data developed here enable systematic comparison of alternative DAC representation in future system assessments.