Global variable resolution modeling using unstructured grids has emerged during the last decade as a downscaling tool to provide high-resolution simulations over regions of interest. Numerical experiments using global variable resolution atmosphere models with prescribed sea surface temperature have demonstrated their effective use in reproducing the behaviors of global uniform high-resolution simulations in the regions of interest at a fraction of the computational cost. These efforts motivated explorations of the global variable resolution modeling approach for coupled climate simulations. Recently, the Energy Exascale Earth System Model (E3SM) featuring a nonhydrostatic atmosphere model on the cubed sphere grid coupled to the Model for Prediction Across Scales Ocean Model (MPAS-O) on the Voronoi grid has been used to perform coupled climate simulations at variable resolutions. In the North American Regional Refined Model (NARRM), the atmosphere model is configured at ~25 km grid spacing over North America and the surrounding areas and ~100 km elsewhere while the ocean model is configured at grid spacings ranging from 14-60 km, with higher resolutions in the oceans surrounding North America. Coupled simulations using NARRM show improved skill in simulating the North America regional climate at higher resolution compared to the uniform low-resolution counterpart. Extending the variable resolution modeling approach to kilometer-scale modeling, the Simple Cloud Resolving E3SM Atmosphere Model (SCREAM) has been used to perform variable resolution simulations with regional refinement down to a few kilometer resolutions in regions of interest. Run in a weather hindcast mode covering periods of days to weeks, these simulations have demonstrated useful skill in simulating extreme weather events. Using examples of coupled and uncoupled regional refined simulations at weather to climate timescales, this presentation will discuss the benefits of global variable resolution modeling as well as remaining challenges.