SLO-based alerting
Every hop emits traces, error rates, and latency histograms. Alerts page the on-call engineer when an SLO burn rate exceeds budget.
Reliability
/
incident history
Every incident,
in the open.
Timelines, impact, and postmortems for every disruption to the KToken gateway, edge network, dashboard, and inference fleet. We publish updates while an incident is active — not after it is over.
90-day uptime
99.98%
Across all components
Incidents
5
Last 90 days
Median resolution
42 min
From detection to resolved
Active now
1
Being monitored
Planned work
Scheduled maintenance
2 upcoming windows
Add to your calendar
Last 90 days
Incident history
All times UTC
Timelines updated live
Showing 5 incidents
How we work
Incident response
On-call in 3 regions
Fifteen-minute ack
The on-call engineer acknowledges within 15 minutes, declares severity, and opens a public incident with the first customer-facing update.
Updates while active
We post on this page and email subscribers at each state change: investigating, identified, monitoring, and resolved.
Blameless postmortems
Incidents above 15 minutes get a written postmortem with root cause, resolution, and tracked preventive actions.
Hear about it from us first
Status emails go out when an incident opens, when the cause is identified, and when it is resolved.