Model Status

Availability of every model, measured from real customer traffic over the last 90 days.

Uptime comes from real requests, not synthetic probes. Customer-caused errors such as no balance or an invalid request are not counted. Status reflects the most recent requests to each model.