Service level agreement
What we commit to, how we measure it, and what happens when we fall short.
Availability
We target 99.99% availability per calendar month for cloud and GPU instances, measured at the network boundary of the region. Scheduled maintenance is announced at least 48 hours in advance and excluded from the calculation.
Support response
Production incidents receive a first human response within 90 seconds, 24 hours a day. Non-urgent tickets are answered within one business day. Growth service requests are answered within one business day.
Service credits
If monthly availability falls below 99.99%, you may request a credit: 10% of the monthly fee below 99.99%, 25% below 99.9%, and 50% below 99%. Credits are applied to the next invoice and must be requested within 30 days of the affected month.
Exclusions
The SLA does not cover outages caused by customer application code, third-party networks outside our control, force majeure, or accounts suspended for violation of the acceptable use terms.
Need the full contract?
Enterprise customers can request a signed SLA and data processing agreement.
Request documents