One whole NVIDIA DGX Spark in EU-Central, billed per minute from prepaid credit.
Connecting to GPUwerk…
Checking capacity…
Pricing & billing · Setup guides · Service status
Need help choosing? Discuss your workload.
This is the offline demo of console.gpuwerk.com. Everything below runs from this one file, state lives in your browser localStorage, and nothing here reaches a real server.
EU-Central · checking availability · first top-up: pay $10, get $20 in credit
Every instance is one whole machine. No other renter shares it.
Unified memory is the whole point, a single node holds models that would otherwise need several discrete GPUs.
Per-minute billing, $0 setup fee, no egress charges. Stopping holds your Spark at a discount off the hourly rate; terminating is what stops billing. First top-up: pay $10, get $20 in credit. Refer a friend: when their first top-up earns the bonus, you get the same amount, up to $10.
Every instance is a fresh container on its own machine, with a hostname and key-based SSH. You are root inside your container, not on the host; bring your own stack. Managed model serving is coming soon.
# deploy from the API curl -X POST https://api.gpuwerk.com/v1/instances \ -H "Authorization: Bearer $GPUWERK_TOKEN" \ -d '{"image":"dgx-os","ssh_public_key":"ssh-ed25519 AAAA..."}' # your shell in your container ssh [email protected] # the whole GB10 is yours nvidia-smi && python3 --version # locked yourself out? reinstall every key on the account, no rebuild needed curl -X POST https://api.gpuwerk.com/v1/instances/<id>/reset-access \ -H "Authorization: Bearer $GPUWERK_TOKEN"
The short version of how billing and instances work.
Per-minute, against a prepaid credit balance. Stopping an instance holds its Spark for you at a discount off the hourly rate; terminating releases the hardware and is what actually stops billing. Your first top-up of at least $10 earns a matching bonus of up to $10 (pay $10, get $20 in credit). If your balance runs out, your instances are stopped and released automatically and each workspace is saved for 7 days, then deleted; nothing is silently overdrafted.
Stopping keeps everything on the node's NVMe, start it again and your files, models, and environment are exactly as you left them. Terminating releases the hardware and permanently erases local data, which is why it requires typing the instance name to confirm.
No. The hourly rate is the whole price, bandwidth in and out over the 200 Gb/s ConnectX-7 link is included.
Each instance is one whole physical DGX Spark, no virtualization, no shared GPUs, no noisy neighbors. SSH access is key-based only; password login for SSH is disabled across the fleet. Optional VNC desktop access, if you enable it at deploy time, uses a separately generated password for that connection only. Hardware is located in EU-Central.
Not yet, today every instance is a clean box you drive over SSH: install vLLM, SGLang, or whatever stack you prefer, and it all fits in unified memory. Preloaded model images that serve on boot are on the roadmap.
First top-up: pay $10, get $20 in credit. Refer a friend: when their first top-up earns the bonus, you get the same amount, up to $10. Model preparation can take several minutes. $0 setup fee, no commitments, no egress charges.
Your instances, credit and hosting at a glance.
Start, stop, connect to and terminate the machines you rent.
Choose the hardware, then a prebuilt image. Access and the running cost come last.
Reserve dedicated Sparks for 1 to 24 months; terms of 3 months or more are discounted.
Your first month is held on your card to secure it and charged once your dedicated Sparks are provisioned, within one business day. If you need more Sparks than we have racked, your first month is charged immediately instead, and we deliver within 14 days or refund it in full. You authorise your card once and each following month is charged automatically.
The last 200 lines your container wrote to stdout and stderr. Stopping shuts the container down and reserves this Spark for you at a discount off the hourly rate, so there is no new output until you start it again; nothing leaves the machine.
| Time | Event |
|---|
Your credit, spending, budget, invoices and billing details. Credit is prepaid; it does not reserve a machine.
Add prepaid credit now so it is ready when you deploy. Credit does not reserve a Spark or start an instance; live capacity is shown when you configure.
Demo mode, credits are added instantly without payment. First top-up: pay $10, get $20 in credit (we match up to $10). Refer a friend: when their first top-up earns the bonus, you get the same amount, up to $10.
When your balance drops below $25, is added automatically, billed at a 10% discount.
Every movement on your prepaid balance, newest first.
| Date | Description | Amount |
|---|---|---|
Share your link. When someone you refer makes their first qualifying top-up, you earn a matching credit too.
| Date | Status | Credit |
|---|---|---|
GB10 machines you bought in the GPUwerk store: order status, shipping, invoices, and the machines you own.
Loading…
Authenticate against the GPUwerk API (api.gpuwerk.com/v1) to deploy and manage instances programmatically.
| Name | Key | Scopes | Created | Status |
|---|
Public keys installed on your instances at deploy time. Print yours with cat ~/.ssh/id_ed25519.pub, or create one with ssh-keygen -t ed25519.
Keys are saved to your account, so they follow you to another browser or machine. A key is written into an instance's authorized_keys when that instance is created and nothing rewrites it afterwards, so deleting a key here stops it being offered for future deploys, but whoever holds it can still log in to instances already running. Rebuild or terminate those to actually revoke access.
| Name | Fingerprint | Added |
|---|
Take a copy of everything we hold about your account, or close the account for good.
One JSON file: your account details, your full billing ledger and payment history, your instances and their event log, your SSH keys, and every message we have sent you. API key values are not included: we only ever store their hashes. Instance workspaces are not part of this export: a running or stopped instance's workspace lives on its node.
Loading…
A small team running real hardware: you talk to the people who run it.
We answer as quickly as we can during business hours. We don't publish a guaranteed response time or uptime figure; where one is agreed it lives in a separate written service level agreement.
Clusters larger than Reserve offers, or a setup that is not in the console yet: email us and we will tell you what is possible and when. To rent out a Spark you own, use Host your Spark.