OBSERVED BY ZROUTER
Provider availability.
Measured from here.
Scheduled inference checks show whether selected AI models respond through ZRouter. These observations describe our connection to providers, not their worldwide service status.
Availability history
Last 30 daysLoading availability observations…
Green: operational · Amber: degraded · Red: outage · Gray: unknown
Percentages cover completed, eligible checks. Missing checks are unknown and are never counted as successful. Current status needs at least three recent observations. Authentication and request configuration errors are excluded from availability percentages.
HOW TO READ THIS PAGE
Useful signals.
Clear limits.
What does ZRouter check?
A small, uncached inference request to each configured target model. Checks normally run every five minutes. A working model listing alone does not count as a successful inference check.
Why might a provider show unknown?
There may be too few checks, no configured target, a monitoring interruption, or observations that are too old. Unknown does not mean a provider is healthy or down.
Does this page include my requests?
No. Public history uses synthetic probes only. It contains no customer prompts, responses, API keys, account identifiers, or traffic volumes.
Where can I see request performance?
Your dashboard contains your usage and request statistics. Operators can also inspect provider attempt history, first-token latency, response duration and generation speed in the overview.