llamaperf

H200 vs RTX 3090

For running local LLMs · 79 reports across 12 models

Side A
H200
Vendor
nvidia
VRAM
141GB
Memory
Discrete
Side B
RTX 3090
Vendor
nvidia
VRAM
24GB
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 →

ModelH200RTX 3090
Qwen3.8up to 2400B382.0n=45
Qwen3.6up to 35B104.0n=12
DeepSeek V4 Flashup to 285B4.8n=260.0n=9
GLM-5.2up to 744B7.3n=2
Qwen2.5up to 32B70.0n=2
DeepSeek V3n=1
DeepSeek V4up to 284B15.0n=1
GLM-4.5-Airup to 106Bn=1
GLM-5.355.0n=1
HobbyLMup to 1Bn=1
Museup to 30B94.0n=1
Qwen3up to 235B7.5n=1

More comparisons