Comment by DiabloD3

Thats not a meaningful question. Models can be quantized to fit into much smaller memory requirements, and not all MoE layers (in MoE models) have to be offloaded to VRAM to maintain performance.