How renting a GPU server works
Short answers to the questions people ask before their first order.
1. Choose and pay
Pick a machine in the catalog, choose a day, a week or a month, and pay from your balance. Top up the balance in the dashboard first. The price you see is what you pay — it's fixed for the whole period.
2. Get access
Access is by SSH key. Paste your public key when ordering (it's saved for next time). Starting a machine takes a few minutes while the software image downloads; then the dashboard shows the SSH command to connect.
You get root inside an isolated container with direct access to the GPUs and NVIDIA drivers already installed.
3. Where the machines are
The machines run in data centers in many countries. Each one in the catalog shows its country, network speed and measured reliability — we only list hosts with at least 98% reliability. Prices change with availability, which is why the catalog updates every few minutes.
4. Extending
Extend from the dashboard at any time before the period ends. We email you a day before it ends. A machine can be extended as long as its host keeps offering it — the catalog shows the maximum number of days for each machine.
5. When the period ends
When the paid period ends, the server and its disk are deleted and can't be recovered. Keep your data in your own storage and copy results off in time.
6. If something goes wrong
If a machine fails to start or stops working during a paid period, write to support from the dashboard or by email — see Contacts.