OpenAI's official status page tracked the incident through several stages. The company listed the incident as "Investigating" for roughly four hours before applying a mitigation fix . After applying the fix, OpenAI moved to "Monitoring" status to track recovery. By the time 9to5Mac reported on the outage, the incident had been ongoing for about four hours, with a mitigation applied roughly two hours prior
. User reports on Downdetector began trending downward after the fix was applied, though some users on social media still reported issues
.
A separate incident on the same day tracked "Image generation unavailable in ChatGPT and API" through its own investigation, mitigation, and monitoring phases, indicating that multiple issues were being addressed concurrently .
The July 23 outage was not an isolated event. OpenAI's systems had experienced several disruptions in the weeks leading up to it:
On the same day as the outage, OpenAI executed a large, pre-announced wave of model deprecations. On July 23, 2026, OpenAI shut down approximately 15 model snapshots and aliases, including:
computer-use-preview-2025-03-11gpt-5-chat-latest, gpt-5-codexgpt-5.1-chat-latest, gpt-5.1-codexgpt-5.2-codexgpt-4o-search-previewo3-deep-research, o4-mini-deep-researchDevelopers were directed to migrate to newer models such as gpt-5.5, gpt-5.4-mini, and gpt-5.5-pro . These removals were announced months in advance — OpenAI had notified developers on April 22, 2026 — but because they overlapped with the unplanned outage, many users and developers likely faced difficulty distinguishing between error messages caused by the service disruption and errors caused by the sudden removal of familiar model aliases from the API
.