The short version: DeployGPU lets you rent GPUs from several clouds with one account and one prepaid balance, and it is run by Quanta Cloud LLC, the same company behind GpuPerHour, GPUIndexes and the QuantaCloud GPU cloud. This page explains what that means for you, including the parts that are not finished yet.
What DeployGPU is
DeployGPU is a marketplace and a control panel in one. It lists live GPU offers from QuantaCloud, Massed Compute, Verda and Hyperstack as virtual machines, plus Vast.ai containers in the dashboard, labelled testing only. You pick an offer, launch it as a plain GPU VM or with one of three ready apps (JupyterLab with PyTorch, ComfyUI, or Open WebUI with Ollama), and connect over SSH or through a private link. The same VM launches are available through a REST API.
The provider accounts behind every launch belong to DeployGPU. You never create an account at Verda or Hyperstack, never hand over your own cloud credentials, and never juggle five different bills. You add credits to DeployGPU, and DeployGPU pays the providers.
Where the prices come from
When I was building GpuPerHour, which runs scrapers against about 30 cloud providers continuously, I learned that a GPU price is only useful with a timestamp next to it: stock and rates move by the hour. DeployGPU uses the same scrapers. Every provider it launches on is checked every minute, and the pages on this site refresh their live tables every 15 minutes. That is why the GPU pages show a timestamp next to every table: a price without a time is a guess.
GpuPerHour still covers the whole market, including clouds DeployGPU cannot launch on. When you want to compare, GpuPerHour is the place. When you want to launch, DeployGPU is.
How the money works
You pay the hourly rate shown on each offer. Any platform fee is already inside that rate, and the deploy page breaks it down before you confirm. Credits are bought by card, with a processing fee of 5% + $0.30 per top-up. The first hour is reserved when you launch, unused seconds come back when you stop, and a launch that fails is refunded automatically. The pricing page has the full mechanics and worked examples.
Who runs it
DeployGPU is operated by Quanta Cloud LLC, which also runs GpuPerHour, the GPUIndexes rental price indices and the QuantaCloud GPU cloud. That last one matters for you: QuantaCloud is one of the providers you can launch on here, so one of the clouds in the marketplace belongs to the same company as the marketplace itself. I would rather say that plainly on this page than have you find it out later.
I am Faiz Ahmed, and I write most of the guides on this site. Where a guide describes something that went wrong on the platform, such as the Open WebUI sign-in loop we fixed on 24 September 2026, it is because it really happened and the fix is live.
What DeployGPU does not do yet
A platform is easier to trust when it is clear about its edges. This is what DeployGPU does not do today:
- Launch a cluster of several VMs, or connect machines over InfiniBand. The largest self-serve machine is one VM with 8 GPUs, and larger needs go through a quote request.
- Pause an instance so that billing stops, or snapshot it. Stopping deletes the instance and its disk. Verda and Hyperstack VMs can be powered off and on through the API, but billing continues while they are off.
- Stop idle instances for you, or cap how long one runs.
- Share an app link with a teammate, or put it on your own domain.
- Issue invoices. You get receipts and a CSV export instead.
- Offer an uptime SLA, spot pricing, or reserved pricing you can buy yourself.
The pages on this site describe the platform as it works today, and each one shows the date it was last updated.
Getting in touch
Questions about an account, a bill or a deployment go to support@deploygpu.ai, and everything else to the contact page. If you are planning more than 8 GPUs or a fixed term, the cluster request form reaches the same team.