Service Level Agreement
Availability commitment and service credits for paid dedicated endpoints. Effective 5 September 2026.
1. Scope
This SLA applies to paid dedicated endpoints operated by Deep Inference under an Order. It does not apply to design-partner, trial, beta or “coming soon” features, or to deployments running in the customer’s own cloud account, where availability depends on infrastructure the customer controls. For in-cloud deployments, response and remediation commitments are set in the Order.
2. Commitment
Deep Inference will make each dedicated endpoint available with a Monthly Uptime Percentage of at least 99.9%.
3. Definitions
- Monthly Uptime Percentage is the total minutes in the calendar month, minus Downtime minutes, divided by total minutes, excluding Scheduled Maintenance and Excluded Events.
- Downtime is any one-minute interval in which more than 5% of well-formed requests to the endpoint return a 5xx error or receive no response within 60 seconds, as measured by our monitoring and confirmed against your logs on request.
- Scheduled Maintenance is maintenance announced at least 48 hours in advance, limited to four hours per month, and scheduled outside your stated peak window where possible. Optimization roll-outs behind the endpoint are performed without downtime and do not count as maintenance.
4. Service credits
| Monthly Uptime Percentage | Credit (% of that endpoint’s monthly fee) |
|---|---|
| Below 99.9% and at or above 99.5% | 10% |
| Below 99.5% and at or above 99.0% | 25% |
| Below 99.0% | 50% |
Credits are capped at 50% of the affected endpoint’s fee for the month, are applied to a future invoice, and are not refundable as cash. Credits are the sole and exclusive remedy for any failure to meet this SLA.
5. Claiming a credit
Email hello@deepinference.co within 30 days of the end of the month in which the Downtime occurred, with the endpoint id, the dates and times of the incident, and request logs or error rates from your side. We will confirm or dispute the claim within 10 business days.
6. Exclusions
Downtime does not include unavailability caused by: factors outside our reasonable control, including internet, DNS or upstream cloud-provider failures; your applications, equipment or network; requests that exceed agreed rate or concurrency limits; suspension for breach of the Terms or non-payment; Scheduled Maintenance; or configurations you froze or set against our written advice.
7. Changes
We may update this SLA with 30 days’ notice. Changes that reduce the commitment do not apply to an Order until its renewal.
This SLA is part of the Terms of Service.