OpenAI incident: Errors processing large PDFs in the Responses API

1. resolvedOct 6, 2026, 17:50 UTC

   All impacted services have now fully recovered.

2. identifiedOct 6, 2026, 15:29 UTC

   We are implementing the mitigation.

3. investigatingOct 6, 2026, 15:23 UTC

   We are investigating the issue for the listed services.

What APIStatus.watch saw

• Oct 6, 2026, 15:48 UTCoperational → degraded

Other services with incidents at the same time

Overlapping incidents aren't necessarily related.

• Elasticsearch: Delayed Serverless Metrics in Azure UAE North regionMajor · Oct 6, 2026, 14:04 UTC
• Snowflake: INC20000263Major · Oct 6, 2026, 14:44 UTC
• CircleCI: Delays in pipelines, workflows, UI data, and notificationsMajor · Oct 6, 2026, 16:00 UTC
• Vercel: Vercel Auth login failures for pages with Deployment Protection enabledMajor · Oct 6, 2026, 16:10 UTC
• Telnyx: Delayed delivery of 10DLC SMS messages to AT&T destinationsMajor · Oct 6, 2026, 17:38 UTC
• Cursor: Investigating service degradation — Security Reviewer AgentsMajor · Oct 6, 2026, 17:49 UTC
• Cloudflare: Cloudflare One Clients are incorrectly challenged on some sitesMinor · Sep 22, 2026, 19:23 UTC
• Zoom: Service Degradation Affecting Customer Using Polycom Devices Going OfflineMinor · Sep 25, 2026, 17:45 UTC
• Twilio: MMS Delivery Delays from a Subset of Twilio Long Codes to GCI United StatesMinor · Sep 26, 2026, 03:16 UTC
• Clerk: Development instances only - Apple rejecting emailsMinor · Sep 28, 2026, 15:23 UTC

How often does OpenAI have incidents?

OpenAI reported 28 incidents in the last 90 days, 5 of them major or critical. A typical incident lasted 55 min. See OpenAI's uptime and incident history