OpenAI Outage History

OpenAI users experienced degraded performance with Codex Cloud Tasks, with follow-up prompts failing at a higher than expected rate. The issue affected Codex / Codex Web workflows that rely on Cloud Tasks, causing errors and delays for those tasks. Status updates noted the degradation began at 3:11 PM UTC and the service was restored by 4:26 PM UTC, with recovery occurring after a little over an hour.

Users calling the gpt-5.5 model via the OpenAI API experienced increased errors and higher latency, affecting the APIs /Chat Completions and /Responses endpoints. The disruption began at 12:31 PM UTC when OpenAI flagged a degraded state and announced an investigation.

OpenAI experienced degraded transcription capabilities that affected ChatGPT and Codex across multiple interfaces, including Codex Web, App, API, CLI, VS Code extension, and ChatGPT Voice mode. Users reported errors loading and transcribing audio, or receiving delayed or incomplete responses in transcription workflows, with conversations looping or stalling in some cases. The impact was observed across regions, with reports from Lima, Peru; Toronto, Canada; Spokane, USA; and Buenos Aires, Argentina. The incident began around 5:16 PM UTC and persisted for several hours as OpenAI worked on mitigation and recovery.

OpenAI's image generation endpoints under the APIs / Images component experienced an elevated error rate beginning around 7:57 AM UTC, causing some image-generation requests to fail or time out for a subset of users. The issue affected developers relying on the image features of the OpenAI API and led to degraded ability to generate images during the incident window.

OpenAI's ChatGPT Apps experienced elevated errors starting at 1:15 PM UTC, affecting a subset of apps and integrations and causing degraded responses for some users. The issue was acknowledged by OpenAI at Feb 26, 2026 2:07 PM UTC and mitigation was implemented by Feb 26, 2026 3:31 PM UTC. Service recovery followed, with OpenAI reporting restoration by Feb 26, 2026 4:02 PM UTC, and the incident lasting roughly 2 hours 47 minutes from first reports to full restoration.

During the incident, OpenAI issued 3 status updates detailing the issue, mitigation steps, and ongoing monitoring. The impact was concentrated on the ChatGPT / Connectors/Apps component, with affected users experiencing intermittent failures within the affected apps and connectors while other OpenAI services remained unaffected.

OpenAI reported a temporary outage affecting artifact generation (creating spreadsheets or slides) on ChatGPT for all users. The incident impacted 13 ChatGPT components, including Conversations, Login, Compliance API, Search, File uploads, Voice mode, GPTs, Image Generation, Deep Research, Agent, Codex, ChatGPT Atlas, and Connectors/Apps, causing users to be unable to generate artifacts during the window. The outage lasted about 1 hour, beginning at Feb 25, 2026 8:32 PM UTC and resolving by Feb 25, 2026 9:32 PM UTC. The issue appeared to affect ChatGPT services globally rather than being region-specific. The provider issued two status updates during the incident, detailing the investigation and restoration of affected services.

OpenAI reported an elevated error rate impacting ChatGPT Conversations for Business and Enterprise customers, affecting the ability to start and maintain conversations. The issue began at 3:36 PM UTC and lasted about 35 minutes, with users experiencing errors and slower responses before recovery. The incident was acknowledged on the OpenAI status page and resolved within a short window.

OpenAI experienced degraded performance in ChatGPT responses for a subset of users, resulting in elevated latency for ChatGPT / Conversations. The issue affected user experience with slower reply times, lasting from 10:01 PM UTC to 10:32 PM UTC (about 31 minutes). OpenAI issued three status updates during the incident, including an initial alert of increased latency, a follow-up investigation for affected services, and a confirmation that latency issues had been resolved. The impact appeared limited to ChatGPT / Conversations and did not reflect a full service outage.

OpenAI’s Sora 2 video generation service experienced degraded performance, with users reporting errors and delays when trying to generate videos. The incident began around 2:45 PM UTC and the OpenAI status page flagged degraded performance at 3:36 PM UTC, as engineers began work on a mitigation. A further update followed at 4:16 PM UTC as the issue persisted, and the service returned to normal by 4:31 PM UTC. The outage lasted roughly 1 hour and 46 minutes.

OpenAI experienced a brief outage affecting the Shopping Research feature within ChatGPT / Search. The outage began at 8:51 PM UTC and was resolved by 9:17 PM UTC. The incident disrupted the ability to use the Shopping Research capability, impacting user workflows that rely on product search and discovery. OpenAI documented the issue with three status updates and applied mitigation measures during the recovery process. The problem appeared isolated to the Shopping Research component (ChatGPT / Search) and did not indicate wider service disruption across OpenAI products. User reports were minimal, indicating limited widespread impact.