← AI Hashrate 中文

RTX 5090 32GB vs V100 32GB SXM2 for Local LLMs

The RTX 5090 32GB has about 99% more memory bandwidth (1,792 vs 900 GB/s). Batch-1 decode tracks bandwidth closely, so a model that fits both cards should decode roughly 99% faster on the RTX 5090 32GB. Both fit the same number of catalog models at Q4/8K (43 of 73; 32 vs 32 GB VRAM). MSRP is $1,999 for the RTX 5090 32GB 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: 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 RTX 5090 32GB posts the higher tok/s in 56 rows vs 1. Methodology.

Spec comparison

RTX 5090 32GBV100 32GB SXM2
VendorNVIDIANVIDIA
VRAM32 GB32 GB
Memory typeGDDR7HBM2
Memory bandwidth1,792 GB/s900 GB/s
TDP575 W300 W
MSRP (list)$1,999$10,000
BuySearch Amazon · eBay · Newegg · B&H Photo

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)RTX 5090 32GB tok/sRTX 5090 32GB fitsV100 32GB SXM2 tok/sV100 32GB SXM2 fitsWinner
MiniCPM5-1B1.081800.7 est.Yes895.6 est.YesRTX 5090 32GB
Qwen3.5-2B2.0972.4 est.Yes483.6 est.YesRTX 5090 32GB
DeepSeek-R1-0528-Qwen3-8B-layer-mix-bpw-3.8-mlx2.3845.6 est.Yes420.5 est.YesRTX 5090 32GB
Qwen3.6-35B-A3B36.0648.3 est.Yes322.4 est.YesRTX 5090 32GB
Qwen3.5-35B-A3B34.7648.3 est.Yes322.4 est.YesRTX 5090 32GB
GLM-4.7-Flash31.2648.3 est.Yes322.4 est.YesRTX 5090 32GB
Qwen3-30B-A3B-Instruct-250730.5648.3 est.Yes322.4 est.YesRTX 5090 32GB
Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF1630.0648.3 est.Yes322.4 est.YesRTX 5090 32GB
Ornith-1.0-35B35.0648.3 est.Yes322.4 est.YesRTX 5090 32GB
qwen35b-a3b-fable-sft-abliterated34.7648.3 est.Yes322.4 est.YesRTX 5090 32GB
Llama-3.2-3B-Instruct3.2607.8 est.Yes302.3 est.YesRTX 5090 32GB
Qwen3-Coder-Next79.7534.2 est.No268.3 est.NoRTX 5090 32GB
Qwen3-Next-80B-A3B-Instruct81.3514.2 est.No258.2 est.NoRTX 5090 32GB
gemma-4-26B-A4B-it25.2511.8 est.Yes254.5 est.YesRTX 5090 32GB
Qwen3-4B-Instruct-25074.0486.2 est.Yes241.8 est.YesRTX 5090 32GB

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

RTX 5090 32GBV100 32GB SXM2
Catalog models that fit43 of 7343 of 73
Largest exclusive fit

RTX 5090 32GB details → · V100 32GB SXM2 details →