OpenAI Outage History

OpenAI experienced an elevated error rate affecting ChatGPT and the API, resulting in degraded performance for users. The incident began at 12:02 PM UTC and was resolved by 12:46 PM UTC, lasting about 44 minutes. A sizable volume of user reports accompanied the issue, indicating broad impact across environments using OpenAI services.

The OpenAI status page posted two updates during the incident, first noting ongoing investigation and then confirming resolution as services returned to normal.

OpenAI's ChatGPT website and API experienced degraded performance due to bad gateway errors beginning around 11:46 UTC. Users reported 502/503 gateway errors and login issues across the ChatGPT web and mobile interfaces as well as the API. OpenAI acknowledged the issue at 12:02 UTC, noting an elevated error rate for ChatGPT and API and that investigation was ongoing. The incident prompted two official updates before the status page returned to a stable state at 12:46 UTC. StatusGator detected the issue at Jan 23, 2025 11:46 AM UTC, about 16 minutes before OpenAI's acknowledgement at Jan 23, 2025 12:02 PM UTC. Some users continued to report issues through Jan 23, 2025 2:17 PM UTC.

OpenAI's Embeddings API experienced high error rates beginning at Jan 21, 2025 8:56 AM UTC, causing failed or degraded embedding requests for some users. The Embeddings API component was affected, leading to slower responses and intermittent failures for embedding generation. The incident lasted about 11 hours and 11 minutes, ending at Jan 21, 2025 8:07 PM UTC when the service returned to normal operation. OpenAI issued 2 status updates during the incident as they worked on a fix and the Embeddings API moved from warn to up at the end of the window.

OpenAI reported a brief degradation in performance impacting the 4o and 4o-mini models accessed through the API. Users may have experienced higher error rates or slower responses during the incident. The outage began at 3:57 AM UTC and was resolved by 4:17 AM UTC, lasting about 20 minutes. OpenAI issued 2 status updates during the event, first indicating degraded performance and ongoing investigation, then confirming the issue was resolved and services returned to normal. The impact was limited to the API component, with no other services noted as affected.

The OpenAI DALL-E API experienced degraded performance due to a high error rate that affected image generation. Users attempting to generate images may have encountered failures or delays while the issue was ongoing. The incident impacted the DALL-E API component and was acknowledged on OpenAI's status page as a degraded service.

The outage began at 10:56 PM UTC and was resolved by 11:37 PM UTC, lasting about 41 minutes. OpenAI issued two status updates during the incident, confirming ongoing investigation and subsequent resolution.

OpenAI reported a scattered network issue beginning Jan 15, 2025 4:08 PM UTC that could impact ChatGPT users. The issue was described as scattered across different areas and led to a downgraded experience for some users, with reports of login loops and unresponsive ChatGPT interactions indicating broader access and interaction problems.

OpenAI issued two status updates as the incident unfolded. The status moved from warn to up at Jan 15, 2025 6:21 PM UTC, signaling restoration. The outage lasted approximately 2 hours and 13 minutes. Reports came from multiple regions, including Costa Rica and Tennessee, suggesting broad but not universal impact.

OpenAI reported a brief outage affecting ChatGPT's web browsing capability, limiting access to up-to-date information. The incident began at Jan 14, 2025 11:01 AM UTC and was resolved by Jan 14, 2025 11:21 AM UTC, lasting roughly 20 minutes. OpenAI posted two status updates during the event, first indicating investigators were looking into the issue and later confirming the service was restored. The degradation was limited to the web browsing feature; core chat functionality remained available for most users.

OpenAI experienced a brief degradation of the Vision API for the gpt-4o-2024-05-13 model, resulting in higher error rates and degraded image-processing performance for users. The incident began at Jan 8, 2025 7:46 PM UTC and was resolved by Jan 8, 2025 8:01 PM UTC, lasting 15 minutes. OpenAI posted two status updates during the incident, noting degraded performance and then the resolution. The issue affected the API component and was limited to Vision capabilities, with no broader impact reported.