DGX Spark (GH200 480GB) vs V100 16GB PCIe for Local LLMs
The V100 16GB PCIe has about 80% more memory bandwidth (900 vs 500 GB/s). Batch-1 decode tracks bandwidth closely, so a model that fits both cards should decode roughly 80% faster on the V100 16GB PCIe. The DGX Spark (GH200 480GB) fits more of the catalog at Q4/8K — 70 vs 25 of 73 models (480 vs 16 GB VRAM). Largest model that fits the DGX Spark (GH200 480GB) but not the V100 16GB PCIe: DeepSeek-V4-Pro (861.6B). MSRP is $3,000 for the DGX Spark (GH200 480GB) vs $8,000 for the V100 16GB PCIe. MSRP is a launch list price, not live retail — Amazon prices move; click through for the current price. Not retail-shoppable: V100 16GB PCIe (datacenter-class; cloud rental is the realistic way to use it). In the Q4/8K head-to-head below (49 shared models), the V100 16GB PCIe posts the higher tok/s in 43 rows vs 6. Methodology.