Future climate projections are subject to multiple sources of uncertainty. Here, we examine model uncertainty from a storyline perspective using three variable-resolution (VR) models: the Community Earth System Model, the Icosahedral Nonhydrostatic model, and the Model for Prediction Across Scales. We implement the storyline framework by prescribing sea surface temperatures (SSTs) and sea-ice concentrations (SICs) from models from the coupled model intercomparison project phase six previously deemed representative of contrasting Arctic and Antarctic storylines. We perform experiments with and without polar refinement, resulting in the first storyline-based multi-model experiments with polar refinement. </p> <p>We evaluate whether this framework captures known characteristics of the storylines and generally find good agreement between the VR experiments and storylines for near-surface temperature, reflecting the strong constraint from the lower-boundary forcing. For precipitation and zonal wind, the characteristics are better captured for the Antarctic storyline experiments than the Arctic ones. Nevertheless, the contrasting storyline forcing mostly yields distinct future responses, demonstrating that storyline-based forcing-model selection is a powerful tool for generating contrasting model experiments. </p> <p>We also quantify the influence of the storyline, VR model, and resolution for a wider range of variables. Resolution has the weakest overall impact, while the storyline dominates across most variables in the Antarctic storyline experiments and for low-level tropospheric temperature and near-surface humidity in the Arctic experiments. The model and resolution influence are generally strongest for precipitation and cloud cover. Thus, the choice of lower-boundary forcing and model has a greater influence on future responses than local changes in horizontal resolution.