Service Level Agreement
Version 1.1.0—Last updated: 2026-10-07
Published
1. Uptime Commitments
- Developer (formerly Explorer): no uptime commitment; the Developer plan is provided on a best-effort basis.
- Qpher Vault Organization: no uptime commitment and no service credits. We work to keep the Service available, and handle incidents as described in Sections 6 and 7.
- Institution: the uptime commitment and any service credits are set out in its agreement, measured as described in Section 2.
2. Uptime Measurement
Uptime is calculated monthly using the following formula:
Uptime % = (Total Minutes in Month - Downtime Minutes) / Total Minutes in Month x 100
Downtime is defined as any period during which the API Gateway (api.qpher.ai) returns HTTP 5xx errors or is unreachable for more than one (1) consecutive minute, as measured by Qpher's external uptime monitoring. Intermittent errors lasting less than one consecutive minute are not counted as downtime.
3. Exclusions
The following events are excluded from uptime calculations and do not constitute downtime for SLA purposes:
- Planned Maintenance: Scheduled maintenance announced by email to affected customers at least 72 hours in advance. Planned maintenance windows are limited to a maximum of 4 hours per month.
- Force Majeure: Events beyond Qpher's reasonable control, including but not limited to natural disasters, acts of government, internet backbone failures, and widespread infrastructure outages affecting major cloud providers.
- Customer-Caused Issues: Downtime caused by the Customer's misuse of the Service, exceeding the rate limits, hard limits or fair-use pools in the Acceptable Use Policy (Section 3), including a reduced rate applied under its Section 4, or issues with Customer-side infrastructure.
- Third-Party Failures: Outages of third-party services (Stripe, SendGrid, etc.) that do not affect core cryptographic API functionality.
- Beta or Preview Features: Any features explicitly designated as beta, preview, or experimental.
4. Service Credits
Service credits are available only under an Institution agreement, as set out in it. Except where applicable law provides otherwise, where an Institution agreement provides service credits they are the sole remedy for failing to meet its uptime commitment.
5. Latency Targets
Qpher targets the following p95 (95th percentile) end-to-end API response latency for warm instances:
- KEM Encrypt: < 150ms
- KEM Decrypt: < 200ms
- Signature Sign: < 250ms
- Signature Verify: < 100ms
End-to-end latency includes the full request lifecycle: API Gateway authentication, policy evaluation, inter-service routing, cryptographic operation, key storage access, and audit logging.
Our targets for the post-quantum operations themselves are under 15ms for ML-KEM-768 encapsulation and under 30ms for ML-DSA-65 signing. The majority of end-to-end latency is attributable to TLS overhead between microservices, Cloud KMS decryption of private keys, and secure key storage access.
Latency may vary based on payload size, network conditions, geographic distance from the us-east1 infrastructure region, and whether the request requires private key access (decrypt, sign) versus public key only (encrypt, verify). Cold-start latency (first request after an idle period) may be significantly higher due to container initialization. Latency targets are goals, not guarantees, and are not covered by SLA credits.
6. Support Response Times
Support response times vary by plan and issue severity:
Developer: documentation; email to support@qpher.ai is answered as capacity allows. No target response time.
Qpher Vault Organization: email support. Target response within 1 business day.
Qpher Vault Organization with 50 or more seats: priority support. Target response within 4 business hours for service-affecting issues and 1 business day for other inquiries.
Institution: as set out in its agreement.
Business hours are Monday through Friday, 9:00 AM to 6:00 PM US Eastern Time, excluding US federal holidays. Support response times are targets, not contractual commitments, except where an Institution agreement says otherwise.
7. Reporting and Transparency
During incidents that affect core API services, we aim to email the owners of affected organizations, and we provide an incident summary on request. Qpher reviews its service targets quarterly and may adjust them with 30 days' notice.
8. SLA Effective Date
This version of the Service Level Agreement is effective as of October 7, 2026. For new customers it applies from the start of their paid subscription; trial periods are not covered.
Qpher LLC
Registered in the Commonwealth of Virginia, United States.
8401 Mayland Dr Ste A, Richmond, VA 23294, USA