GPU · FPGA · on demand
Bare-metal accelerators, rented by the minute.
Pick a machine from live inventory, SSH in as root, install your own stack. Prepaid credit, per-minute billing, no contracts.
Instances
Sign in to see live availability and hourly rates.
- ›Dedicated GPU, not time-sliced
- ›Root SSH on a blank Linux VM
- ›NVMe scratch, wiped on return
- ›Dedicated accelerator card
- ›Bring your own toolchain and license
- ›Program the card from your own VM
- ›Multiple cards, one tenant
- ›Longer-running workloads
- ›Volume pricing on request
How it works
1 · Top up credit
Prepaid balance, no commitment. Spend it across any instance type, down to the minute.
2 · Pick a machine
Browse live availability, pick the accelerator and spec you need, and boot a blank Linux machine.
3 · SSH in
Root access over SSH from minute one. Install your own toolchain, run your own licenses.
Built for builders
Root SSH, always
Every instance is a full Linux machine you own while it runs. No shared runtimes, no sandbox walls.
Per-minute metering
Billing starts at boot and stops at destroy. Nothing rounds up to the hour.
Bring your own stack
Blank machines with the accelerator attached. Your tools, your licenses, your kernels.
Predictable pricing
One flat hourly rate per machine, shown up front. What you see is what is deducted.
Request early access
Inventory is opening in waves. Tell us what you're building and we'll send an invite and current pricing.
[email protected]