← AI Hashrate 中文

H200 141GB SXM5 vs V100 32GB SXM2 for Local LLMs

The H200 141GB SXM5 has about 433% more memory bandwidth (4,800 vs 900 GB/s). Batch-1 decode tracks bandwidth closely, so a model that fits both cards should decode roughly 433% faster on the H200 141GB SXM5. The H200 141GB SXM5 fits more of the catalog at Q4/8K — 59 vs 43 of 73 models (141 vs 32 GB VRAM). Largest model that fits the H200 141GB SXM5 but not the V100 32GB SXM2: Qwen3-235B-A22B-Instruct-2507 (235.1B). MSRP is $35,000 for the H200 141GB SXM5 vs $10,000 for the V100 32GB SXM2. MSRP is a launch list price, not live retail — Amazon prices move; click through for the current price. Not retail-shoppable: H200 141GB SXM5, V100 32GB SXM2 (datacenter-class; cloud rental is the realistic way to use it). In the Q4/8K head-to-head below (57 shared models), the H200 141GB SXM5 posts the higher tok/s in 57 rows vs 0. Methodology.

Spec comparison

H200 141GB SXM5V100 32GB SXM2
VendorNVIDIANVIDIA
VRAM141 GB32 GB
Memory typeHBM3eHBM2
Memory bandwidth4,800 GB/s900 GB/s
TDP700 W300 W
MSRP (list)$35,000$10,000
Buy——

MSRP is the launch list price, not live retail. Buy links are Amazon affiliate links — the price after you click is the live price.

Q4 / 8K head-to-head — top 15 of 57 shared models

ModelParams (B)H200 141GB SXM5 tok/sH200 141GB SXM5 fitsV100 32GB SXM2 tok/sV100 32GB SXM2 fitsWinner
MiniCPM5-1B1.084823.4 est.Yes895.6 est.YesH200 141GB SXM5
Qwen3.5-2B2.02604.7 est.Yes483.6 est.YesH200 141GB SXM5
DeepSeek-R1-0528-Qwen3-8B-layer-mix-bpw-3.8-mlx2.32264.9 est.Yes420.5 est.YesH200 141GB SXM5
Qwen3.6-35B-A3B36.01736.4 est.Yes322.4 est.YesH200 141GB SXM5
Qwen3.5-35B-A3B34.71736.4 est.Yes322.4 est.YesH200 141GB SXM5
GLM-4.7-Flash31.21736.4 est.Yes322.4 est.YesH200 141GB SXM5
Qwen3-30B-A3B-Instruct-250730.51736.4 est.Yes322.4 est.YesH200 141GB SXM5
Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF1630.01736.4 est.Yes322.4 est.YesH200 141GB SXM5
Qwen3-Coder-Next79.71736.4 est.Yes268.3 est.NoH200 141GB SXM5
Ornith-1.0-35B35.01736.4 est.Yes322.4 est.YesH200 141GB SXM5
Qwen3-Next-80B-A3B-Instruct81.31736.4 est.Yes258.2 est.NoH200 141GB SXM5
qwen35b-a3b-fable-sft-abliterated34.71736.4 est.Yes322.4 est.YesH200 141GB SXM5
Llama-3.2-3B-Instruct3.21627.9 est.Yes302.3 est.YesH200 141GB SXM5
gpt-oss-20b21.51447.0 est.Yes268.7 est.YesH200 141GB SXM5
gemma-4-26B-A4B-it25.21370.9 est.Yes254.5 est.YesH200 141GB SXM5

est. = bandwidth-based estimate; measured = curated real benchmark that overrides the estimate. Fits = weights + KV(8K) + 1 GB overhead ≤ 95% of VRAM.

Fit coverage at Q4 / 8K

H200 141GB SXM5V100 32GB SXM2
Catalog models that fit59 of 7343 of 73
Largest exclusive fitQwen3-235B-A22B-Instruct-2507 (235.1B)—

H200 141GB SXM5 details → · V100 32GB SXM2 details →