OpenAI Outage History

OpenAI reported increasing error rates in Codex at 3:56 PM UTC, with the Codex-powered components experiencing issues. Users attempting Codex-based workflows, including tasks in ChatGPT / Codex, may have seen errors or degraded behavior. As the incident evolved, related components ChatGPT / Deep Research and ChatGPT / Login were also affected.

OpenAI Codex on the web was unavailable for all users starting at May 27, 2025 9:41 PM UTC. The issue affected Codex features accessed through the OpenAI web interface, rendering Codex unusable for everyone. OpenAI listed the ChatGPT / Codex component as affected and stated they were investigating the cause.

The outage lasted until May 27, 2025 10:52 PM UTC, roughly 1 hour and 11 minutes. OpenAI issued 2 status updates during the incident as they worked to restore access to Codex on the web, with services returning to normal after the restoration.

OpenAI experienced a billing-system issue that left some Pro and Team subscribers unable to access their subscriptions. The outage began on May 26, 2025 6:47 PM UTC and was resolved by May 27, 2025 5:28 PM UTC, lasting about 22 hours and 41 minutes. The problem affected a broad set of ChatGPT components, including ChatGPT / Login, ChatGPT / Subscriptions, ChatGPT / Conversations, ChatGPT / Memory, and other features, leading to disrupted access to paid subscription capabilities for affected users. User reports described login and API-page errors during the incident. The incident involved multiple official updates as the team worked to diagnose and restore access, with the status moving from warn to up as services came back online.

OpenAI Codex task creation and PR creation encountered errors, affecting developers attempting to create new Codex tasks or pull requests via Codex workflows. Users likely saw failures or elevated error rates when initiating these actions, with the issue reportedly connected to a degradation in the GitHub API.

The incident began at May 26, 2025 8:11 AM UTC when the OpenAI status page showed the ChatGPT / Codex component moving from a healthy state to a degraded state. The issue was resolved by May 26, 2025 9:46 AM UTC, lasting approximately 1 hour and 35 minutes. OpenAI issued two status updates during the incident, describing elevated errors in Codex task creation and PR creation and noting a potential link to GitHub API issues. Services returned to normal after the final update.

OpenAI reported a brief degradation affecting the Codex component within ChatGPT, with elevated error rates observed for a short period. The incident began at May 23, 2025 6:36 PM UTC and was resolved by May 23, 2025 6:42 PM UTC, lasting about 6 minutes. OpenAI indicated mitigation efforts were underway and the status transitioned from up to warn and back to up as the issue was mitigated.

OpenAI ChatGPT Web and mobile experiences were affected by elevated error rates starting on May 22, 2025 9:41 AM UTC. The incident impacted ChatGPT Web, iOS, and Android interfaces, causing degraded performance for users attempting to interact with the service. OpenAI's status page noted Elevated error rates and indicated the team was investigating.

The issue persisted until May 22, 2025 12:27 PM UTC, when the Web and mobile components returned to normal. OpenAI issued two status updates during the incident, documenting ongoing investigations and the eventual recovery. User reports described API errors such as 500 and 503 responses, consistent with API access problems. The outage lasted approximately 2 hours and 46 minutes and affected users across Web and mobile clients.

OpenAI Codex task processing experienced a brief degraded performance with elevated error rates affecting Codex-related tasks. The issue started at May 22, 2025 1:16 AM UTC and was resolved by May 22, 2025 1:36 AM UTC, lasting about 20 minutes. OpenAI published two status updates during the incident noting high error rates and ongoing investigation, with services returning to normal after the window. The impact was limited to Codex processing and did not constitute a full service outage.

OpenAI reported a brief login issue affecting ChatGPT authentication. Users were unable to sign in to ChatGPT during the incident, impacting access to services requiring authentication. The problem began at May 21, 2025 2:41 AM UTC and was resolved by May 21, 2025 3:51 AM UTC, lasting roughly 1 hour and 10 minutes. The affected area was ChatGPT / Login, and the provider issued two official updates as mitigation efforts progressed.

OpenAI experienced a brief degradation in ChatGPT's Voice mode due to increased errors in voice transcription, impacting users who rely on voice input. The issue began at May 20, 2025 10:01 AM UTC and was resolved by May 20, 2025 10:12 AM UTC, lasting about 11 minutes. A small number of user reports indicated transcription issues around 9:30 AM UTC.

OpenAI's status page documented the incident with updates as the team investigated and then confirmed resolution. The issue affected the ChatGPT / Voice mode component; other ChatGPT features were not reported as impacted, and the status transitions reflected a move from degraded performance back to normal as the incident concluded.

OpenAI's ChatGPT Search experienced a brief degradation in performance, with users encountering elevated failures and high latency when using the search feature. The outage began at May 17, 2025 1:26 PM UTC and was resolved by May 17, 2025 1:42 PM UTC, lasting 16 minutes. OpenAI issued two status updates during the incident, noting an initial investigation and a subsequent restoration of the search service. The impact was confined to the ChatGPT Search component and affected user experiences related to search functionality.