
Facts from Exoscale’s own pricing and docs pages, September 2026 — 30 read from a quoted sentence on the page we archived.
Exoscale rents GPUs from $0.74 per GPU-hour on demand, bills per second, and asks no minimum commitment. Exoscale does not publish an uptime SLA or a refund policy.
| On-demand GPU instances | Volume commitment | |
|---|---|---|
| From | Flat per-second rate, identical in every zone | A discount against on-demand |
| Minimum term | None — no up-front cost, no long-term commitment | A defined minimum commitment, billed whether used or not |
| Billing | Per second, flat rate across all zones | Discounted against a committed level |
| Tenancy | Physically isolated GPUs, single to multi-GPU | Same resources under a usage commitment |
| Provisioning | Scale from a single GPU to multi-GPU without orchestration | — |
As of September 2026, the lowest published on-demand rate is $0.74 per GPU-hour for the NVIDIA A30 24GB. Exoscale publishes an on-demand rate for 6 GPU models. Billing is per second.
Exoscale does not publish an uptime commitment on its own site.
No. Exoscale states none on-demand; volume plans set one deliberately.
Exoscale states its egress terms as outbound transfer charged; rates on the calculator.
Exoscale lists european zones including Zurich (CH-DK-2) and Frankfurt (DE-FRA-1).
The contracting entity is Akenes SA.
Reviewed September 2026, from Exoscale’s own site. 30 facts, each quoted from the page it was read on.