System status

Live health for every published model, based on API requests completed in the last hour.

Torve AI API · Last 60 minutes

Checking recent API traffic

Loading eligible request outcomes from the last hour…

Status is based on completed API request outcomesTraffic evidence, not a synthetic uptime check

How this status is calculated

We monitor eligible customer API requests to the models we publish. Completed calls count as successful; upstream/provider errors—including provider authentication failures and rate limits—count as service failures. Invalid requests, customer quota limits, disconnects and unfinished streams are excluded.

At least five observed requests are needed to classify a window: 99%+ success is operational, below 99% is degraded, and zero successes is a disruption. Fewer than five requests means limited evidence; no requests means no traffic, not verified uptime. Each model’s strip breaks the last hour into twelve five-minute intervals.

This reflects real API-provider traffic, not an uptime guarantee or an active probe of every upstream. Read our availability policy.