All systems operational

• API

  Accepting and routing inference requests.

  Operational
  90-day uptime100.00%p95 end-to-end latency of inference requests in the window—Checks528,533

• Model providers

  Upstream model hosts the gateway forwards to.

  Operational
  90-day uptime100.00%p95 upstream call latency across all providers—Checks143,075

• Media storage

  Delivery of generated images, video and audio.

  Operational
  90-day uptime99.97%round trip of a HEAD on the bucket225msChecks114,130

• Payments & billing

  Top-ups, and how quickly they land as credit.

  Operational
  90-day uptime99.99%median time from callback received to processed—Checks226,631

Uptime is successful checks over total checks, not an estimate. A component that has never been sampled shows — rather than 100%. The same figures are available as JSON from GET /v1/status, which keeps answering during maintenance.