GPU compute · Hourly
GPU compute by the second, paid from a balance in pounds.
Top up a balance, request a GPU from your panel and pay only for the minutes it runs. Our team confirms each request during support hours (Monday to Friday, 08:00 to 18:00 UK time), and billing starts only when the GPU is running. Machines run in EU data centres, you connect over SSH, and everything stops automatically when your balance runs out, so a forgotten job cannot run up a bill.
- spec sheet
- updated 22 October 2026
- Billing
- Per second, from a prepaid balance
- Currency
- Pounds sterling, no VAT added
- Minimum top up
- £25
- Location
- EU data centres only
- Access
- SSH, from your panel
- Start
- Requests confirmed Mon to Fri, 08:00 to 18:00 UK
- Safety net
- Automatic stop at zero balance
How billing works
- 01
Create an account and top up
Add at least £25 to your balance by card. Your first top-up is checked by a person before you can start GPUs; this keeps fraud off the platform.
- 02
Request a GPU
The panel shows the GPU types available right now and the per-second price of each, in pounds. Send a request and our team confirms it during support hours (Monday to Friday, 08:00 to 18:00 UK time); we email you when the GPU is starting. Nothing is charged while a request waits, and you can cancel it.
- 03
Run your job
Once the GPU is running, connect over SSH and work as you would on any Linux machine. Usage is deducted from your balance minute by minute, from the moment it is running.
- 04
Stop when you are done
Stop the GPU from the panel. If your balance runs low we email you; if it reaches zero the GPU stops automatically. Starting a stopped GPU again may also need a confirmation from our team; the panel tells you when it does.
Prices change with GPU availability, so we show the current per-second rate in the panel rather than printing a table here that could be out of date.
What people use it for
- Fine-tuning or evaluating open-weight models without buying hardware.
- Batch inference, transcription and embedding jobs that run for hours, not months.
- Rendering, simulation and research workloads with a clear start and finish.
- Trying a GPU type before committing to a Private AI server.
Cryptocurrency mining and anything that breaks our acceptable use policy is not allowed.
Limits that protect you
- A cap on how many GPUs one account can run at the same time, raised on request once you have a history with us.
- A spending limit per account, so a mistake in a script cannot drain a large balance in one go.
- Automatic stop when your balance runs out, so spending is limited to what you have topped up.
Price
Hourly GPU
Pay as you go
Prepaid balance, billed per second.
Per-second rate shown in the panel for each GPU type. Minimum top up £25.
- Billed per second, only while running
- Requests confirmed in UK support hours
- EU data centres
- Automatic stop when the balance runs out
- SSH access
Prices in pounds. No VAT is added: we are not VAT registered.
FAQ
Questions we get asked
How much does a GPU cost per hour?
It depends on the GPU type and current availability. Your panel shows the current per-second price in pounds for every available GPU before you send a request, and that price is fixed for the GPU you request.
How quickly can I start a GPU?
You can send a request from your panel at any time. Our team confirms GPU requests during support hours, Monday to Friday, 08:00 to 18:00 UK time, and we email you when the GPU is starting. You are not charged until it is running, and you can cancel the request before then. A request that is not confirmed in time expires, and we tell you.
Where do the GPUs run?
In EU data centres only. We do not run hourly GPUs outside the EU.
What is the minimum top up?
£25, paid by card. You can top up more at any time.
Can I keep a GPU running permanently?
You can, but if you need one around the clock a Private AI server at a fixed monthly price is usually better value and comes managed.
Read next
- GPU computePrivate AI server: your own GPU running open models, managed for youA dedicated GPU server running open-weight AI models for your team, with model serving, monitoring and updates handled for you. Fixed monthly price by GPU type, quoted before setup.
- GuidesRunning a private AI model in a UK firm: a UK GDPR checklistWhat to check before you run an open-weight AI model on your own server for client or staff data: lawful basis, DPIAs, processors, transfers, model licences and security, with links to ICO guidance.
- PricingPrices for managed servers, migrations, AWS and GPUEvery WharfCloud product and price in pounds, from managed servers to GPU compute. No VAT is added.
Ready when you are.
Order online, or ask an engineer first if you want to check the fit. Nothing is set up until you have paid, and every first order is checked by a person.