Status
Real-time check of public endpoints from your browser. Live status reflects whatever your browser can reach right now (subject to your own network conditions).
Live endpoint check
Checks run when you load this page. Refresh to re-check. If you see all green and your application still fails, the issue is likely client-side (check your API key, base URL, network).
SLA expectations during alpha
| Tier | Commitment |
|---|---|
| Availability | Best-effort. Realistic 99.5%-99.8% historically. |
| Latency (chat) | p50 under 5s, p95 under 15s on epithre-omni for typical 500-token replies. |
| Latency (embed) | p50 under 500ms for batch-of-64. |
| Incident notification | Email to all registered customers within 1 hour of detection. |
| Credit for downtime | Not currently offered. We'll work it out case-by-case if needed. |
We expect to formalize these into a contractual SLA when we move from alpha to GA. Email hello@epithre.com if you need an SLA contract today.
Maintenance windows
We do not have a fixed maintenance window during alpha. Most upgrades are zero-downtime (rolling restart of stateless gateway). Backend model swaps may cause 1-2 minute outages on specific endpoints, announced 24 hours ahead in #announcements (email).
Recent incidents
No incidents in the last 30 days.
When incidents happen they get logged here with date, duration, affected endpoints, root cause, and remediation. To get notified by email when an incident is opened or resolved, send a message to hello@epithre.com with subject "Subscribe incidents".
Reporting an incident
If you observe an outage or degradation that isn't reflected on this page:
- Capture evidence:
X-Request-IDheader, error JSON, timestamp. - Email
hello@epithre.comwith subject "Suspected incident". - We acknowledge within 1 hour during business hours, 4 hours off-hours.
Subscribe to status updates
- Email:
hello@epithre.comwith subject "Subscribe status" - we'll add you to the announcement list. - Webhook
platform.status_changeevent: on the roadmap.
Infrastructure footprint
- Primary datacenter: Jakarta, Indonesia. All inference and storage runs in-country.
- CDN/edge: not yet deployed. Customers worldwide hit Jakarta directly.
- Backups: scheduled daily, off-site. Restore tested monthly.
See also
- Changelog - recent platform updates.
- Troubleshooting - common issues + fixes.
- Rate limits