OpenAI Outage History

OpenAI experienced broad elevated error rates across multiple services, notably Sora Video generation and a wide range of ChatGPT features. Users reported server errors, timeouts, sign-in problems, and difficulty loading conversations or histories. The disruption appeared global and persisted for roughly two hours, beginning around Jul 16, 2025 1:36 AM UTC.

OpenAI's status page documented the investigation and mitigation across several updates. The provider issued seven updates between Jul 16, 2025 1:36 AM UTC and Jul 16, 2025 3:51 AM UTC, noting mitigation steps and ongoing monitoring as recovery progressed. By the final updates, many components had returned to normal, with recovery completed for the majority of services by around Jul 16, 2025 3:51 AM UTC.

Scope and impact: The incident affected more than a dozen ChatGPT subsystems in addition to Sora Video generation, including Web, Codex, Context connectors, Desktop and Mobile clients, Memory, File uploads, Dall-E, Login, Conversations, and related features. User reports tracked a global pattern of issues across regions, consistent with a widespread outage rather than a regional fault. Overall, the event represented a major disruption to OpenAI services during the window.

OpenAI experienced a period of elevated error rates affecting ChatGPT and Sora video generation, beginning around 1:36 AM UTC. Users reported sign-in failures, conversations not loading, and failed or slow requests across ChatGPT features and Sora. The disruption appeared global, with reports of server errors and timeouts across multiple regions as a broad set of ChatGPT components and Sora were affected for roughly two hours, impacting access to chats, content generation, and related functionality.

OpenAI's ChatGPT and API services experienced elevated error rates that affected a broad range of features and API endpoints. User reports began around 4:30 PM UTC, with OpenAI acknowledging the issue at 5:02 PM UTC and indicating mitigation efforts. The incident involved dozens of ChatGPT components and API paths, including login, messaging, file uploads, search, image generation, embeddings, and more, leading to degraded performance and failed requests for users. The disruption lasted about 7:46 PM UTC, with services gradually recovering by that time.

The OpenAI 4.1-mini API experienced elevated errors and degraded performance starting at 9:31 PM UTC, with a broad degradation across 15 API components including Assistants, Audio, Images, Fine-tuning, Realtime, Responses, Login, Batch, Embeddings, Files, Completions (legacy), Chat, Compliance, Moderations, and Uploads. Users attempting requests to the 4.1-mini API observed errors or slower responses, impacting typical workflows.

OpenAI issued two status updates describing the degradation and ongoing mitigation, and the incident was resolved by 9:57 PM UTC. The outage lasted about 26 minutes. User reports were limited but indicated intermittent issues across the affected services.

OpenAI experienced a partial outage affecting multiple API endpoints, including /v1/chat/completions, /v1/completions, the Private Link, and the API responses surface. The status page indicated an issue labeled as Errors on API Endpoints, with the affected components listed as APIs / Responses and APIs / Completions (legacy). The incident involved two status updates from OpenAI during the event, and normal operation was restored after.

OpenAI experienced an outage impacting the API used to create new fine-tuning jobs. Users attempting to start fine-tuning runs encountered errors, disrupting workflows that rely on programmatic model training. The incident affected the APIs / Fine-tuning component and appeared to be global based on reports from multiple regions.

Timeline: The outage began at Jul 8, 2025 12:16 AM UTC and was mitigated by Jul 8, 2025 1:22 AM UTC. The status subsequently moved to a degraded state and recovery was completed by Jul 8, 2025 6:56 PM UTC. OpenAI issued 3 official status updates during the incident.

Impact and recovery: The outage lasted approximately 18 hours and 40 minutes, with full fine-tuning job creation restored by late Jul 8, 2025 6:56 PM UTC. User reports indicate global distribution of issues, confirming widespread impact. OpenAI's team implemented mitigation and monitoring as the service returned to normal.

OpenAI's ChatGPT Codex experienced elevated error rates, leading to degraded performance for users. The incident began at 9:36 PM UTC and was resolved by 10:17 PM UTC. OpenAI issued two status updates during the event, first noting elevated errors and that a mitigation was being implemented, then confirming the service had returned to normal. The issue affected the ChatGPT / Codex component and caused noticeable disruption for a brief period.

OpenAI experienced a brief degradation across its API endpoints beginning at Jul 7, 2025 6:01 PM UTC and lasting about 36 minutes until Jul 7, 2025 6:37 PM UTC. The disruption affected a broad set of API components, causing requests to fail or time out for developers and their applications. Affected components included APIs / Realtime, APIs / Responses, APIs / Login, APIs / Fine-tuning, APIs / Assistants, APIs / Audio, APIs / Batch, APIs / Embeddings, APIs / Files, APIs / Images, APIs / Completions (legacy), APIs / Chat, APIs / Compliance, APIs / Moderations, and APIs / Uploads. OpenAI issued two status updates during the incident to communicate the elevated error rates and ongoing mitigation.

OpenAI's updates also noted that users were experiencing elevated errors across the impacted endpoints and that mitigation steps were underway, with services returning to normal by the end of the window.

OpenAI Pay-As-You-Go API accounts experienced a temporary quota-related incident, with some users seeing 429 errors indicating quota issues. The issue affected 15 API components, including Chat, Completions, Embeddings, Images, Audio, and more, causing requests to fail for affected accounts. Users reported messages such as 'You exceeded your current quota' during API calls. The incident involved multiple API surfaces and degraded the experience for pay-as-you-go users.

OpenAI's Sora service experienced a brief outage that affected multiple Sora components, including Image Generation, Video Generation, Video Viewing, and Feed. Users reported elevated errors and partial access issues during the event. The outage began at Jul 1, 2025 4:47 PM UTC and was resolved by Jul 1, 2025 5:11 PM UTC, lasting about 24 minutes.

OpenAI's status page documented two updates during the incident. The first noted increased errors for Sora and identified the affected components, while the second confirmed restoration of services. The incident involved four Sora components and was reflected in user reports from across regions.