Electrochemical conversion of CO2 to fuels and valuable products is one pathway to reduce CO2 emissions. Electrolyzers using gas diffusion electrodes (GDEs) show much higher current densities than aqueous phase electrolyzers, yet models for multi-physical transport remain relatively undeveloped, often relying on volume-averaged approximations. Many physical phenomena interact inside the GDE, which is a multiphase environment (gaseous reactants and products, liquid electrolyte, and solid catalyst), and a multiscale problem, where ‘‘pore-scale’’ phenomena affect observations at the ‘‘macro-scale’’. We present a direct (not volume-averaged) pore-level transport model featuring a liquid electrolyte domain and a gaseous domain coupled at the liquid–gas interface. Transport is resolved, in 2D, around individual nanoparticles comprising the catalyst layer, including the electric double layer and steric effects. The GDE behavior at the pore-level is studied in detail under various idealized catalyst geometries configurations, showing how the catalyst layer thickness, roughness, and liquid wetting behavior all contribute to (or restrict) the transport necessary for CO2 reduction. The analysis identifies several pathways to enhance GDE performance, opening the possibility for increasing the current density by an order of magnitude or more. The results also suggest that the typical liquid–gas interface in the GDE of experimental demonstrations form a filled front rather than a wetting film, the electrochemical reaction is not taking place at a triple-phase boundary but rather a thicker zone around the triple-phase boundary, the solubility reduction at high electrolyte concentrations is an important contributor to transport limitations, and there is considerable heterogeneity in the use of the catalyst. The model allows unprecedented visualization of the transport dynamics inside the GDE across multiple length scales, making it a key step forward on the path to understanding and enhancing GDEs for electrochemical CO2 reduction.