OpenAI Outage History

OpenAI reported elevated errors affecting Codex and the Responses API, causing degraded performance for users relying on code generation and API responses. The incident began at Dec 17, 2025 1:06 AM UTC and impacted the ChatGPT / Codex and APIs / Responses components. OpenAI issued two status updates during the incident, noting that a mitigation had been applied and they were monitoring the recovery.

The outage lasted about 3 hours and 10 minutes and was resolved by Dec 17, 2025 4:16 AM UTC, with services returning to normal.

OpenAI Codex CLI and its Responses API experienced elevated errors beginning at 6:46 PM UTC, affecting two components: ChatGPT / Codex and APIs / Responses. Users saw degraded performance and errors when using Codex CLI, ChatGPT, and API calls. The outage lasted until 12:41 AM UTC, about 5 hours and 55 minutes.

OpenAI's Image API experienced a short-lived degradation starting at 4:16 AM UTC, with a high error rate affecting image-related requests. Users may have encountered errors or degraded image generation/processing during this period. The issue was mitigated and services returned to normal rapidly after a brief interruption. A single user report appeared around the incident window, indicating some users were impacted.

OpenAI experienced a widespread outage starting at Dec 9, 2025 10:51 PM UTC that impacted ChatGPT connectors and a broad set of core ChatGPT components. Users encountered connector disconnects and degraded experiences across Conversations, Login, Compliance API, Search, Codex, Deep Research, Agent, File uploads, Voice mode, GPTs, Image Generation, and ChatGPT Atlas. The outage produced errors and slow responses, with prompts failing to load and repeated reconnect attempts reported by users around the world. The incident persisted with staged recoveries over several days as services gradually returned to normal for different components.

OpenAI published seven official status updates during the incident, describing the root cause as ChatGPT connectors disconnecting and noting that a fix had been rolled out but full restoration would take additional time. Updates were issued across the period from 2025-12-10 through 2025-12-12 as connectivity and affected components recovered at different paces. By Dec 12, 2025 11:17 PM UTC all major components were restored and users were advised that some apps might require reconnecting affected integrations.

Geographically, reports appeared globally with affected users across multiple regions and devices. The outage lasted approximately 3 days, from Dec 9, 2025 10:51 PM UTC to Dec 12, 2025 11:17 PM UTC, with the service returning to normal in a staged fashion as the connector system stabilized.

OpenAI experienced a partial outage impacting Codex Web features and Code Reviews used with ChatGPT. Users reported failures and degraded performance when attempting to access Codex Web tooling and code-review workflows. The issue appeared regionally, with reports from Europe (United Kingdom and The Netherlands).

The outage began at 8:26 PM UTC and was resolved by 9:21 PM UTC, lasting about 55 minutes. OpenAI issued two status updates during the incident, acknowledging the investigation and restoration of the affected components. The disruption affected Codex Web tooling and code-review workflows for a portion of users, rather than a full platform outage.

OpenAI experienced a brief disruption affecting ChatGPT Voice mode, with users reporting elevated errors and login difficulties. The issue began at Dec 9, 2025 7:36 PM UTC and was resolved by Dec 9, 2025 8:11 PM UTC, lasting about 35 minutes. A small number of user reports surfaced during the incident, including a post from Fishers, Indiana noting login failures, which aligned with the degraded voice features.

From Dec 8, 2025 11:32 PM UTC the OpenAI status page indicated an authentication issue tied to GitHub credentials affecting ChatGPT and Codex Web. Users worldwide reported login failures, blank pages when asking questions, and error messages across multiple devices and browsers. The impact appeared broad, with reports from the United States, United Kingdom, Canada, and Zambia among others.

The provider issued 2 status updates during the incident, first noting the authentication error and later confirming resolution. The status went from a listed issue to up at Dec 9, 2025 5:56 AM UTC, marking a duration of about 6 hours and 24 minutes. The volume and distribution of user reports indicate widespread impact on workflows relying on GitHub-authenticated access to OpenAI ChatGPT and Codex features.

OpenAI reported GitHub authentication errors with Codex Web, causing degraded chat access for a subset of users. Some users experienced issues loading chats, chats stuck in loading, and difficulties deleting conversations. The issue affected the ChatGPT / Codex component and began at 7:41 PM UTC.

OpenAI issued 2 status updates during the incident, and services were restored by 10:11 PM UTC, for a total duration of about 2 hours and 30 minutes. Reports came from multiple regions, indicating broad impact though not universal.

OpenAI experienced a partial disruption affecting two features: the ChatGPT Slack Connector could not be installed for some users, and the CustomGPT UI failed to display the Allow/Deny button. This limited the ability to connect Slack with ChatGPT and to manage custom GPTs for affected users. The incident began at Dec 5, 2025 1:41 AM UTC and was resolved by Dec 5, 2025 4:27 AM UTC, lasting approximately 2 hours and 46 minutes. User reports were limited during the event.

OpenAI issued three status updates during the incident, confirming issues across two components: ChatGPT / Connectors and ChatGPT / GPTs. By Dec 5, 2025 4:27 AM UTC both components had returned to normal operation.

OpenAI experienced a brief outage in the ChatGPT Conversations component, with users reporting elevated errors and partial access issues for roughly 30 minutes. The outage began on 10:26 AM UTC and was resolved by 10:56 AM UTC, during which OpenAI initiated mitigation efforts to restore normal operation.