This page is showing its last published state from ago. That usually means the main deployment cannot publish right now.
Uptime over the past 14 days.
No incidents reported.
Resolved. A standby model took over the affected lane automatically. Requests were served throughout; no customer action was needed.
Investigating. We are seeing reduced capacity from an upstream provider on the Frontier lane.
No incidents reported.
No incidents reported.
No incidents reported.
Resolved. This incident has been resolved. Provider failures of this class now fail over to the requested model automatically.
Monitoring. A fix has been implemented and verified. We are monitoring the lane.
Identified. An upstream provider is degraded. Affected requests are being served on the requested models while substitutions on the lane are paused.
Investigating. We are investigating elevated errors and slow responses on the Workhorse lane.
No incidents reported.
No incidents reported.
No incidents reported.
No incidents reported.
No incidents reported.
Resolved. This issue has been resolved. API traffic was unaffected throughout.
Identified. A deploy regression was identified and rolled back. Challenge rounds are serving again.
Investigating. Challenge rounds are failing. We are investigating.
No incidents reported.
Resolved. Error rates returned to normal. The failure classes involved were addressed in subsequent gateway hardening.
Investigating. We are investigating elevated errors on a portion of API requests.
Resolved. Google and GitHub sign-in are now the primary path and are unaffected. Email link delivery for the affected providers was addressed.
Identified. Some corporate mail security tools consume single-use sign-in links before the recipient can. Affected users saw an invalid-link error.
Investigating. We are investigating reports of email sign-in links arriving already used.