Last updated 1 August 2026. This SLA forms part of the terms of service.
We commit to 99.9% monthly availability for running instances, measured per instance per calendar month. An instance is unavailable during any full minute in which it is powered on, has not been stopped or destroyed by you, and cannot be reached on its public interface or has lost access to its attached GPU.
| Monthly availability | Credit |
|---|---|
| 99.0% to below 99.9% | 10% of that instance's monthly spend |
| 95.0% to below 99.0% | 30% |
| 90.0% to below 95.0% | 60% |
| Below 90.0% | 100% |
Credits are applied to your account balance and can be spent on any service. They are not refundable to a payment method and are capped at your total spend on the affected instance for the month in question.
If a host failure terminates a running instance, we credit the full cost of that instance's runtime for the hour in which it failed, automatically and without a claim. This is separate from and additional to the availability credits above.
The commitment does not cover unavailability caused by:
Availability credits under section 2 require a claim. Email support@gpuai.shop within 30 days of the end of the affected month with the instance IDs and the approximate times of unavailability. We respond within five business days.
Service credits are the only remedy for failure to meet this SLA. Nothing here excludes liability that cannot lawfully be excluded.
We may change this SLA with 30 days' notice by email to account holders. Changes never apply retroactively to a month already in progress.