Verda GPUs on DeployGPU

The Verda offers on DeployGPU are GPU virtual machines in Finland, run by Verda, the GPU cloud that was called DataCrunch until November 2025 (Verda's announcement). DeployGPU launches them on its own Verda account, so you do not need one: you pay from your DeployGPU balance, and you log in over SSH as root.

Offers now

6

From

$0.22/GPU-hr

GPUs per VM

1x, 2x, 8x

GPU models

5

Finland · checked 2026-10-15 19:08 UTC

Verda GPUs you can launch now

Each row below is one GPU family in stock on Verda right now, with the number of VM offers, the GPU counts per VM and the cheapest price per GPU-hour. The list refreshes every 15 minutes, and Verda's stock changes through the day, so a GPU missing here can be back within hours.

The cheapest individual offers, each with a Launch link that opens the deploy page with the full cost breakdown before anything is charged:

DeployGPU only lists what it can launch. For Verda prices next to the rest of the market, see Verda on GpuPerHour.

How a Verda VM differs from the other 3 VM providers

The last row is worth knowing before you launch. DeployGPU keeps one encrypted platform key per account and installs it on Verda VMs, so root access for the platform exists on these machines. Your own key is the one you log in with, and you control it.

Logging in as root on port 22

Verda VMs log you in as root, where QuantaCloud and Hyperstack use ubuntu and Massed Compute uses Ubuntu with a capital U. The deployment page shows the full command:

ssh root@YOUR_VM_IP -p 22

Working as root means no sudo and no second chance: a wrong rm -rf or a broken edit to /etc/ssh/sshd_config takes effect at once. The SSH guide covers keys, and Permission denied (publickey) covers the most common login error.

Disk size, volumes and what survives a stop

Verda is the only VM provider on DeployGPU where you choose the disk size, and one of 2 with volumes (Hyperstack is the other). Pick the size on the deploy page with the model files in mind: Llama 3.3 70B in BF16 is about 141 GB on disk before anything else.

Stopping a Verda deployment deletes the VM and its disk, the same as on every provider. To keep data between sessions, create a volume on the Volumes page before you launch and attach it on the deploy page. A Verda volume lives in Finland, is billed every hour it exists, attached or not, and needs formatting and mounting the first time. Current volume prices are on the pricing page.

Power a Verda VM off and on with the API

Verda is one of 2 providers where DeployGPU can power a VM off and back on without deleting it (Hyperstack is the other). There is no dashboard button for it, and it is not a pause: the deployment stays active and full compute billing continues while the VM is off. It exists to recover a VM that stopped answering, or to power-cycle one while keeping its disk and files.

# Power off, keeping the VM and its disk
curl -X POST https://api.deploygpu.ai/api/v1/deployments/DEPLOYMENT_ID/shutdown \
  -H "X-API-Key: gpu_live_YOUR_KEY"

# Repeat until power_state reports stopped
curl https://api.deploygpu.ai/api/v1/deployments/DEPLOYMENT_ID/power \
  -H "X-API-Key: gpu_live_YOUR_KEY"

# Power the same VM back on
curl -X POST https://api.deploygpu.ai/api/v1/deployments/DEPLOYMENT_ID/start \
  -H "X-API-Key: gpu_live_YOUR_KEY"

Shutdown works only on a running VM and start only on a stopped one. A request for the state the VM is already in returns 200 and does nothing, and an accepted request returns 202. Wait until /power shows the new state before sending the opposite action, and allow a few minutes after a start before SSH answers. The API reference lists every field and error.

What you pay, from the first hour to the last second

The hourly price in the tables is the rate you are billed, and it includes any platform fee. Launching reserves the first hour from your balance, a launch that fails is refunded in full, and when you stop, the unused seconds of the current hour go back to your balance. Card top-ups carry a 5% + $0.30 processing fee. The pricing page explains the rest, and the cost calculator works out a run from live prices.