NVIDIA Tesla V100 32GB (SXM2)

Compare NVIDIA Tesla V100 32GB (SXM2) against another machine in the Arena

Class: GPU
Memory: 32 GB HBM2
Runs (Q4_K_M): —
Bandwidth: 897 GB/s
TDP: 300 W
Released: 2018-03-01
Price (US): $650 used · $709 new as of 2026-08
Price (Canada): CA$893.02 used sourced, as of 2026-08

Sourcing and disambiguation notes

SXM2 module: needs a carrier or server board rather than a plain PCIe slot, as with the V100 16GB. The used price is from gpudojo.com's tracker, and the new figure is new-old-stock from the same tracker; the card is discontinued. The 32GB variant reportedly beats a 24GB RTX 3090 on usable context depth for MoE models, though that article's own benchmark numbers were not well enough sourced to file as records here. The Canadian figure is for the native PCIe card rather than the SXM2 module this entry documents.

Specification sources

https://www.hardware-corner.net/guides/tesla-v100-32gb-for-llm/ — press, 2026-08-13
https://gpudojo.com/tesla-v100 — press, 2026-08-22

Measurements

Decode is token generation — the speed you feel while an answer streams. Prefill is prompt processing — the wait before it starts. Why bandwidth predicts decode speed.

What verified, single-source and estimated mean, and the same rows with every filter and sort in the benchmarks explorer.

Speed over time

What it can run

Fit is arithmetic, not a measurement — how it is computed. Against 32 GB; models that fit are listed largest first.