Skip to content
GPU Finder

4x RTX 5090 cloud pricing

Every 4-GPU RTX 5090 node we track across 4 providers — from $1.18/hr ($0.29/GPU-hr).

Providers:4
4x node:$1.18 — $5.34/hr
From per month:$846.40 (30 d × 24 h)
Spot:from $0.80/hr
In stock:3 providers

Cheapest in-stock 4x RTX 5090 right now

Limited
$1.18/hr$0.29/GPU-hr · $846.40/month
on Vast · Japan, JP
Visit Vast

Pricing by provider

  • 4x RTX 5090 marketplace
    Limited95% · 30d
    $1.18/hr
    4× 4x RTX 509032 vCPU189 GBFree egressJapan, JP
  • 4x RTX 5090 marketplace
    Stale · 10h ago0% · 30d
    $2.00/hr
    4× 4x RTX 509032 vCPU31 GBRussia
  • 4x_56c_476g_RTX5090
    No data
    $2.60/hr
    4× 4x RTX 509056 vCPU476 GBus-east-2, us-east-4 +1 more
  • 4x RTX 5090 Community Cloud
    Limited38% · 30d
    $2.76/hr
    4× 4x RTX 509032 vCPU256 GBFree egressGlobal

RTX 5090 (4× GPU) availability — last 7 days

No provider held 4× RTX 5090 ≥80% available across all 7 days. Polled hourly · 3 providers tracked · 7-day window.

≥80%50–80%<50%No data
View:
Provider
17
18
19
20
21
22
23
Week
Vast94%
Lium66%
Runpod23%

4x RTX 5090 pricing FAQ

How much does a 4x RTX 5090 server cost per hour?

Right now a 4-GPU RTX 5090 node starts at $1.18/hr ($0.29 per GPU-hour) across 4 providers, and the priciest 4x RTX 5090 listing we track is $5.34/hr. Prices are on-demand rates as published by each provider and refresh hourly.

How much does a 4x RTX 5090 cost per month?

At the current floor of $1.18/hr, running a 4x RTX 5090 node around the clock for 30 days costs about $846.40 before egress and storage. Per-second and per-minute billing (see each provider page) avoids paying for partial hours on shorter jobs.

Which provider has 4x RTX 5090 nodes in stock right now?

On our last hourly check the cheapest 4x RTX 5090 node reported in stock was $1.18/hr on Vast (Japan, JP). The table below marks every 4x RTX 5090 listing In stock, Limited, Out or No data, and the heatmap shows each provider's 4x availability over the last 7 days.

RTX 5090 cloud pricing →Cheapest cloud GPU per hour →