OpenAI incident: We are seeing elevated error rates across API models

1. resolvedSep 17, 2026, 21:50 UTC

   All impacted services have now fully recovered.

2. monitoringSep 17, 2026, 21:20 UTC

   We have applied the mitigation and are monitoring the recovery.

3. identifiedSep 17, 2026, 20:20 UTC

   We have identified that users are experiencing elevated errors for the impacted services. We are working on implementing a mitigation.

What APIStatus.watch saw

• Sep 17, 2026, 21:30 UTCoperational → degraded
• Sep 17, 2026, 21:35 UTCdegraded → operational

Other services with incidents at the same time

Overlapping incidents aren't necessarily related.

• Twilio: Voice Call Failures from a Subset of Twilio Phone Numbers to Hong KongMinor · Sep 12, 2026, 00:16 UTC
• Telnyx: SMS Delivery — Encoding Issues on Outbound Messages to T-MobileMinor · Sep 15, 2026, 21:57 UTC
• Notion: Duplicate characters when editing with a Korean keyboard on iOSMinor · Sep 16, 2026, 23:58 UTC
• Replicate: Replicate Pruna issueMinor · Sep 17, 2026, 01:09 UTC
• Notion: Some connected content results unavailable in Notion AIMinor · Sep 17, 2026, 14:07 UTC
• Notion: Notion Calendar sign-in redirects on webMinor · Sep 17, 2026, 15:25 UTC
• GitHub: Elevated rate of errors for OpenAI models provided by CopilotMinor · Sep 17, 2026, 20:59 UTC

How often does OpenAI have incidents?

OpenAI reported 18 incidents in the last 90 days, 4 of them major or critical. A typical incident lasted 52 min. See OpenAI's uptime and incident history