Skip to content
Independent counter-quote desk for GPU clusters, bare-metal fleets & colocation · free for buyers

Server price index

What GPU servers cost when providers actually quote them, next to the public list-price benchmarks. Our series come from quotes we source for buyers, published as anonymous medians once enough providers have quoted.

Quoted prices

Median effective price from quotes we sent to buyers, by GPU, term and region.

No quote-based series has met our publication rule yet: at least 5 quotes from 3 or more providers in 28 days, with no provider above 50% of them. Until a series qualifies, the public benchmarks below are the reference. How the index works

Public list-price benchmarks

Median listed prices across dozens of GPU clouds. Pick a GPU for its history.

GPUMedian reservedMedian on-demand
NVIDIA H100$3.23$3.50
NVIDIA H200$3.68$4.88
NVIDIA B200$6.12$7.99
NVIDIA A100$1.80$1.81
NVIDIA L40S$1.37$1.60

List prices per GPU-hour, USD. Source: GetDeploying, GPU rental price history, CC BY 4.0, checked October 19, 2026 unless marked. Reserved means 12-month reservations, or the nearest term with three or more providers. These are published list prices, not quotes; large reservations usually come in lower.

Is your quote above the index?

Send your spec, invoice or renewal notice. We'll get 3+ providers to bid on it and lay every quote out as one 36-month cost.

Get counter-quotes