Live health of every part of the platform, from the API gateway to upstream provider connectivity.
What each indicator means when you see it on a component.
Normal service.
Elevated latency or partial errors.
Significant impact, a workaround may exist.
Service unavailable.
Scheduled work in progress.
When something breaks, here's how fast we post the first status update and how often we follow up.
| Severity | Definition | First update within | Update cadence |
|---|---|---|---|
| SEV1 | API down or >50% errors | 30 minutes | Every 60 minutes until resolved |
| SEV2 | Major model / provider down | 1 hour | Every 2 hours |
| SEV3 | Degraded performance | 2 hours | As status changes |
Detection: automated monitoring (latency, error rate, uptime), support ticket clusters, and upstream provider status pages. After resolution: a final summary is posted within 24 hours; SEV1 incidents receive a public postmortem within 5 business days.
Get notified the moment a component changes state, and know about planned maintenance ahead of time.
Choose any channel — they all carry the same incident and maintenance notifications.
No maintenance is currently scheduled.
Frontier model routes depend on these providers. We link their status pages in incident updates.