OpenAI mitigates high errors across multiple ChatGPT services – Unite.AI

OpenAI applied a mitigation and was monitoring recovery from high errors affecting multiple ChatGPT products as of early October 6, 2026, according to an incident record on the company’s status page. The record lists five affected components with the incident marked with degraded performance.

The incident record carries the title “Elevated errors affecting multiple products including ChatGPT (conversations), image generations, spaces and pages, API rehydration, and points.” Its first update, published at 2:41 UTC on October 6, 2026, said that OpenAI was investigating the issue for the listed services.

A second update at 2:44 UTC said: “We have applied mitigation and are monitoring recovery.” The incident status is listed as Monitoring with reduced performance. OpenAI’s status history lists the incident as occurring on October 6 with the mitigation update as the latest entry.

Components listed in the incident log

The record lists five affected components, all within the ChatGPT product group: Conversations, ChatGPT Work, Image Generation, Points, and Space. The incident headline separately names Spaces and Pages, Rehydration API, and Dots among the affected products.

OpenAI’s main status page shows the incident as an active banner under a warning that the company is currently experiencing issues, marks it as in progress, and notes that it impacts ChatGPT.

More incidents on the OpenAI status page

The new incident wasn’t the only open point. The main status page showed a second active incident, titled High Error Rates in Codex Cloud, listed under Monitoring Status with Mitigation Applied and marked as ongoing for eight hours. A third incident, titled High Work Mode Errors, was listed as mitigated with recovery under monitoring, marked as ongoing for nineteen hours, and noted as impacting ChatGPT.

The history page shows the most recent Work Mode incident update as of 9:50 AM on October 5, 2026, and the most recent Codex Cloud incident update as of 6:07 PM on October 5, 2026. Two other October 5 entries were resolved: the Admin Console was intermittently inaccessible before OpenAI reported that all affected services had been fully restored at 9:07 PM, and an issue related to Saving and real-time editing in Pages was fixed in an Update at 1:04 PM.

As of early October 6, 2026, the history page lists two more incidents that OpenAI says have been fully resolved: high errors in the Platform Access and Administration APIs, with an update of 1:35, and high errors in ChatGPT conversations, with an update of 12:01.

The log also shows a September 29, 2026 incident involving high errors across ChatGPT, Codex, and APIs, including the Agents API, which OpenAI reported as resolved with all affected services restored. In that article, OpenAI said a detailed root cause analysis would be published within five business days.

Availability data and reporting methodology

The status page reports availability data from July 2026 to October 2026. The ChatGPT group, composed of 16 components, has an uptime of 99.65% for the period and the API group, composed of 13 components, at 99.96%. The Codex Group is traded at 99.95% of four components and the FedRAMP Group at 100% of one component. The Ad Platform group shows two components, with the Ads API having 99.992% uptime.

For the five components named in the new incident, the page reports an uptime of 99.96% for Conversations, 99.86% for ChatGPT Work, 99.79% for Image Generation, 100% for Points, and 100% for Space. Among other members of the ChatGPT group, the page lists Access at 99.97%, Voice Mode at 99.96%, Search at 99.98%, and Connectors/Apps at 99.96%.

Within the API group, the page lists the 99.96% Uptime Images component and the 99.995% Access component. The four components of the Codex group, Codex Web, Codex API, CLI, and the VS Code extension, are each listed with an uptime of 99.95%.

The page states that availability metrics are reported at an aggregate level across all levels, patterns, and error types. It adds that individual customer availability may vary depending on the subscription level, specific model and API features in use.