OpenAI Outage History

OpenAI ChatGPT faced a multi-faceted incident beginning late on Nov 12, 2025 11:31 PM UTC, with enterprise users seeing Access Denied errors, elevated login failures, and Android users experiencing image generation failures. The problems affected ChatGPT Conversations, Login, and Image Generation across regions, hindering enterprise workflows and content creation. The incident persisted through Nov 14, 2025 2:16 AM UTC, lasting roughly 28 hours and 45 minutes.

OpenAI's status page issued six updates during the incident, describing investigations, mitigation efforts, and monitoring as recovery progressed. Milestones included addressing Android image generation and login-related failures, with the Image Generation component returning to normal and overall services resuming by the final update on Nov 14, 2025 2:16 AM UTC.

Reports were globally distributed across teams and regions, with users from North America, Europe, and other areas reporting issues such as access errors, image generation timeouts, and login problems, indicating a broad but not complete outage across ChatGPT features.

OpenAI experienced degraded performance due to latency in the Vector Stores API, impacting vector storage related operations. The APIs / Files component was affected, and users likely noticed slower processing during the incident. The issue began at Nov 12, 2025 5:56 AM UTC and mitigation measures were applied while the team monitored for recovery.

OpenAI posted two status updates during the incident: first moving the service to a degraded state at Nov 12, 2025 5:56 AM UTC, and later restoring normal operation at Nov 12, 2025 8:26 AM UTC. The outage lasted about 2 hours and 30 minutes. Report activity was limited, consistent with a short-lived impact, and OpenAI confirmed recovery on their status page.

OpenAI experienced a partial outage when the Costs API endpoint /v1/organization/cost started returning 404 errors, affecting Login and Chat Completions APIs. As a result, sign-in attempts and chat completion requests could fail or run with degraded performance for a period of time. OpenAI acknowledged the issue and issued updates as they investigated the root cause. The incident affected 2 components: APIs / Login and APIs / Chat Completions, with user reports indicating activity around the affected endpoints.

OpenAI experienced elevated error rates across the Sora API, affecting video generation and video viewing components, beginning at Nov 7, 2025 9:21 PM UTC. Users reported failures generating videos and playback issues, with errors appearing in multiple regions. The disruption also impacted Codex-related services (ChatGPT / Codex), signaling broader impact across OpenAI AI capabilities.

StatusGator recorded an Early Warning Signal for OpenAI when users reported Sora video generation issues beginning around Nov 7, 2025 8:32 PM UTC. The official acknowledgment followed about 49 minutes later, with OpenAI posting the first alert at Nov 7, 2025 9:21 PM UTC noting elevated error rates for the Sora API and that mitigation was being implemented. The problem affected OpenAI's Sora API, impairing video generation and viewing for many users, and there were concurrent elevated errors observed in Codex services as well. OpenAI listed elevated error rates for Sora and indicated mitigation efforts, with affected components including Sora / Video generation and Sora / Video viewing.

OpenAI's ChatGPT service experienced degraded availability for Free users, with elevated errors affecting ChatGPT / Conversations. The issue began at 6:12 PM UTC (10:12 AM PT) and lasted until 7:01 PM UTC (11:01 AM PT), a duration of about 49 minutes.

OpenAI's status page acknowledged the degradation at 6:26 PM UTC and listed ChatGPT / Conversations as degraded with mitigation underway. The incident included two official status updates before the service returned to normal by 7:01 PM UTC.

OpenAI reported a memory-related issue affecting ChatGPT Memory within ChatGPT Conversations beginning at 6:46 PM UTC. Users across multiple regions experienced memory recall failures and errors saving or retrieving memories, impacting how conversations remember past context. The incident manifested as degraded memory functionality within ChatGPT Conversations and was reflected in user reports across a broad base.

OpenAI issued two status updates during the incident. The issue persisted until 4:16 PM UTC, for an outage duration of about 21 hours and 30 minutes. The service was restored after the final update, with ChatGPT Conversations returning to normal operation. Reports spanned North America, Europe, and other regions, indicating widespread but partial impact rather than a full downtime.

OpenAI reported a brief degradation in the ChatGPT / Conversations component as users encountered errors indicating conversations were too long. The degraded period began at Nov 6, 2025 4:11 AM UTC and lasted about 1 hour and 10 minutes, with the service returning to normal by Nov 6, 2025 5:21 AM UTC. The issue affected the ChatGPT conversations flow, causing interruptions for some chats or prompts.

OpenAI's ChatGPT / Codex experienced elevated errors that disrupted generation requests. Users across several U.S. locations reported issues, including 'Please try again' errors and timeouts when generating responses, indicating degraded performance. The incident began at 7:46 PM UTC and lasted about 2 hours and 10 minutes, until 9:56 PM UTC, with 5 user reports observed during the window.

The OpenAI status page indicated an elevated error and that mitigation was being implemented as the ChatGPT / Codex component moved from normal to degraded and back to normal after mitigation. The provider issued 2 official status updates during the incident, and service is now reported as restored.

OpenAI experienced a login incident impacting three login-related components: APIs / Login, Sora / Login, and ChatGPT / Login. Users reported sign-in failures and difficulty accessing affected services. The outage began on 3:02 PM UTC and was resolved by 5:41 PM UTC, lasting roughly 2 hours and 39 minutes. Reports from multiple regions described errors during sign-in and API access attempts.