Simple pricing.
Bigger savings the longer you run.
RTX 5090 from $0.79/hour on demand, billed per minute. 25% off overnight, up to 30% off when you reserve. No storage, egress or setup fees.
Pay the way that fits
Prices for one RTX 5090. Multi-GPU instances (e.g. 2× on one machine) cost the per-GPU price × the number of GPUs.
Automatic for on-demand minutes midnight–8 AM Pacific. active now
See off-peak pricesPrepay 1 day, 1 week or 1 month. The GPU is yours for the whole term.
Reserve a GPURun overnight, pay 25% less
Every on-demand minute that starts between midnight–8 AM Pacific is charged at $0.59/hr instead of $0.79/hr. Nothing to switch on — queue a training run before bed and it's applied minute by minute.
What a job costs
How reserving works
- 1Choose a term1 day (10% off), 1 week (20% off) or 1 month (30% off) — in the GPU console or when you rent.
- 2Paid when it startsThe full amount comes out of your credit once the container is running. If it can't start, you pay nothing.
- 3Yours for the whole termNever stopped for a low balance. At the end of the term it stops and the container is deleted.
- !Early stopStopping early doesn't refund the remaining time. If our host fails, the unused share is credited back.
How billing works
Add credit
Top up your wallet by card, from $20.00. One payment covers many hours.
Start a GPU
You need 60 minutes of credit to start on demand, or the full price to reserve.
Pay per minute
Charged per started minute while running — never while starting, never after you stop.
Stay in control
Auto-stop limits, low-balance warnings, and a full history in your wallet.
Pricing & billing questions
How does billing work?
You add credit to your wallet (minimum $20.00). While a GPU is running, we deduct its hourly price per started minute. You need at least 60 minutes of credit to start a GPU.
When is off-peak, and how does the discount work?
Off-peak is midnight–8 AM Pacific, every day. Each on-demand minute that starts inside that window is charged at 75% of the normal rate. A job running from 10 PM to 10 AM pays full price for 4 hours and the off-peak rate for 8.
How do reservations work?
Pick 1 day, 1 week or 1 month when you rent. The full amount is taken from your credit when the container starts — never if it fails to start. The GPU is yours for the whole term and stops at the end of it. Stopping early doesn't refund the remaining time; if our host fails, the unused part is credited back.
How are multi-GPU instances priced?
The per-GPU price times the number of GPUs — a 2× RTX 5090 instance on demand is $1.58/hr. Off-peak and reservation discounts apply the same way.
What happens when my credit runs out?
We warn you when less than an hour of credit is left. If it reaches zero, your on-demand GPU is stopped and the container deleted. Top up any time to keep running.
Are there storage, bandwidth or setup fees?
No. You pay for GPU time only. The container's disk and internet traffic are included.
Can I get a refund on unused credit?
Yes — contact us and we'll refund your unused balance to the original card. If a rental fails to start, you're never charged for it.
Do you offer invoices for businesses?
Every top-up appears in your wallet history, and enterprise customers are invoiced monthly in CAD.
I need several GPUs or more than a month.
Self-serve reservations go up to one month. For multiple machines, longer terms, guaranteed capacity or an SLA, contact sales for custom pricing.
Start with $20.00.
No contracts. Per-minute billing. Stop whenever you like.