llamaperf

DGX Spark vs H200

For running local LLMs · 26 reports across 10 models

Side A
DGX Spark
Vendor
nvidia
VRAM
128GB
Memory
Unified
Side B
H200
Vendor
nvidia
VRAM
141GB
Memory
Discrete

Highest reported generation speed by model family

These maxima come from different community setups. Model size, quantization, GPU count, offloading and concurrency may differ. They do not establish which GPU is faster for the same workload. Check the comparison conditions →

ModelDGX SparkH200
DeepSeek V4 Flashup to 285B90.0n=104.8n=2
Ling-3.0up to 124B40.9n=3
Qwen3.8up to 2400B181.0n=3
GLM-5.2up to 744B24.0n=2
DeepSeek V3n=1
GLM-5.1n=1
HobbyLMup to 1Bn=1
Kimi K320.0n=1
Museup to 30B37.0n=1
Qwen3.6up to 35Bn=1

More comparisons