OpenAI Outage History

OpenAI experienced a partial outage impacting ChatGPT search and related API endpoints, causing elevated errors and slower responses for prompts that triggered web search and for certain Codex tasks. The outage began at Aug 11, 2025 3:56 PM UTC and was resolved by Aug 11, 2025 10:07 PM UTC, lasting roughly 6 hours and 11 minutes. OpenAI issued four status updates during the incident detailing the progression and the mitigation efforts as engineers worked to restore services. By the time services were restored, the affected search and Codex-related paths were back to normal.

OpenAI's ChatGPT / Conversations experienced a brief outage caused by rate limit or model not found errors in GPT 5, affecting user conversations. The issue lasted about 5 minutes, starting at 7:11 PM UTC and ending at 7:16 PM UTC, with some users encountering errors when interacting with GPT-5 features. OpenAI implemented mitigation and began monitoring the recovery.

OpenAI’s chat and project loading features faced a brief outage, with users reporting chats and projects not loading or responding. The incident began at 5:37 PM UTC and ended at 6:30 PM UTC, lasting about 53 minutes and affecting users globally (reports from North America, Europe, and the Caribbean). StatusGator detected the issue at Aug 8, 2025 5:37 PM UTC, ahead of any official acknowledgment, while OpenAI did not publish status updates during the window.

The OpenAI ChatGPT search feature experienced a partial outage affecting free-tier users. The incident began at 10:46 AM UTC and was resolved by 6:11 PM UTC, lasting roughly 7 hours and 25 minutes. During this window, users reported that searches did not respond or returned errors, with reports coming from multiple regions. The issue impacted the ChatGPT / Search component and likely affected the ability to quickly locate information within ChatGPT.

OpenAI's status page documented the incident under ChatGPT / Search as a degraded service, with 2 official updates issued during the incident. The service was restored to normal after the final update at 6:11 PM UTC. Based on user reports, the impact appeared global rather than region-specific.

OpenAI experienced elevated errors on the ChatGPT 4.5 model starting at Aug 7, 2025 12:31 PM UTC, impacting user interactions with ChatGPT. The incident later expanded to elevated latency and error rates for file uploads to vector stores, affecting APIs / Assistants and APIs / Files. The issue spanned multiple components, including ChatGPT / Conversations and the API endpoints for Assistants and Files, leading to degraded performance and intermittent failures for some users.

OpenAI issued four status updates during the incident. A mitigation was applied and monitoring continued; API/Assistants and API/Files recovered to normal by Aug 8, 2025 12:46 AM UTC, while ChatGPT / Conversations remained degraded until Aug 8, 2025 9:06 AM UTC when it was restored. The incident lasted about 20 hours and 35 minutes and affected users across regions, reflecting a broad impact on OpenAI services.

OpenAI experienced elevated error rates for Deep Research among free users starting at 10:41 AM UTC, with some user reports appearing earlier. The issue affected the ChatGPT / Deep Research component, causing degraded performance for free-tier access. The outage lasted about 1 hour and 25 minutes and was resolved by 12:06 PM UTC.

The provider posted two status updates during the incident, describing the investigation and restoration progress as services returned to normal.

OpenAI's ChatGPT / Conversations component experienced elevated error rates affecting image-based conversations, causing errors or incomplete responses for prompts with images. User reports came from multiple regions, indicating a global impact on image prompts and related chat functionality. The incident began at 8:41 PM UTC and was resolved by 10:01 PM UTC, lasting about 1 hour 20 minutes. OpenAI issued two status updates during the outage, noting ongoing mitigation efforts and eventual restoration. The issue was confined to ChatGPT / Conversations, with degraded performance observed in image-related chats.

OpenAI reported a brief disruption affecting ChatGPT Conversations with images, with elevated errors observed from Aug 6, 2025 7:46 PM UTC to Aug 6, 2025 8:03 PM UTC. A mitigation was applied and recovery was monitored, with the service returning to normal by 8:16 PM UTC. The incident appeared in a small number of user reports and did not indicate a broader outage across ChatGPT. OpenAI provided status updates describing the issue and the recovery process.

An extended outage affected OpenAI's ChatGPT family services, including Codex and a broad set of related components. Users experienced degraded performance, slow responses, and intermittent failures when submitting requests or loading features, with a total of 23 components impacted. The incident began on 10:37 AM UTC and persisted until 7:51 PM UTC, lasting roughly 9 hours and 14 minutes. Reports from multiple regions indicated broad impact across the service, underscoring a widespread disruption rather than a localized issue.

OpenAI issued a series of status updates detailing ongoing mitigation efforts. The provider logged four official updates during the incident, outlining the evolution of the problem and eventual restoration. By 7:51 PM UTC, the affected components listed on the status page had returned to normal, signaling the end of the outage for the majority of users.

OpenAI's ChatGPT conversations for paid users experienced elevated errors for a brief period. The issue affected the ChatGPT / Conversations component, with some users unable to complete conversations or receiving error responses. The outage began at Aug 5, 2025 3:01 PM UTC and was resolved by Aug 5, 2025 3:11 PM UTC, lasting about 10 minutes. OpenAI issued 2 status updates during the incident as they worked on a mitigation.

User reports were limited during the event, suggesting a narrow impact. The status page communications indicated ongoing mitigation efforts and a quick restoration of normal service after the incident window.