In this talk, I will show some recent results on modeling and test-bed development of two vapor-feed device configurations. A multi-physics agglomerate model that describes the operational details for a gas diffusion electrode (GDE) based device, including the local CO2 concentration profiles, pH gradients, polarization losses, CO2 utilization challenges and impact on Faradaic efficiency due to hydrogen evolution reaction, will be discussed in detail. The modeling results were then compared to the performance of Cu nanoparticle based GDEs. As predicted by the model, Cu-GDEs operating at alkaline conditions for CO2 reduction and CO reduction exhibited significantly higher partial current density for C2 products generation than the planar analogy. The transformation of the Cu catalyst in GDE under various electrochemical potentials for CO2 reduction was also investigated by in situ X-ray absorption spectroscopy. In addition, a proof-of-concept, full vapor feed device for CO2 reduction to formic acid using a reverse-configured bipolar membrane to achieve advantageous product separation will be discussed.