Is your AI service down?

Separate an outage from an account or request problem

Identify the failing action: signing in, sending a message, running a task, uploading a file, or receiving an API response. Compare mapped components and incident times on the individual service page. A reachable homepage does not establish that generation or authenticated features work.

For API failures, record the HTTP status and request ID privately. Rate limits and account limits require different handling from server errors. Preserve work before refreshing or repeating a potentially billable operation.

Follow a disruption through recovery

Use the incident radar, save products to your watchlist, or subscribe to incident feeds from their pages. The historical calendar records available evidence and keeps missing coverage visible.