Service Level & Availability

High-level availability posture and how we handle incidents for Evaluator.

Target availability

Evaluator is engineered for high availability with multiple layers of health checks:

/public/healthz – basic uptime and routing.
/public/ready – gateway plus model readiness.
/v1/warm – explicit warm-up control for the decision model.

The Enterprise and Enterprise Pro plans are designed to support demanding workloads. Formal SLAs and credits may be documented in your commercial agreement.

Maintenance windows

We aim to perform maintenance with minimal customer impact. When maintenance is expected to affect availability, we will use the Status page and direct communication for Enterprise/Enterprise Pro customers where applicable.

Incident communication

In the event of a service-impacting incident, we will:

Update the Status page with a brief description and current state.
Work to restore service as quickly as possible, following internal runbooks.
Provide post-incident summaries to affected customers where appropriate.