OpenAI outage records
Read official OpenAI outage and maintenance notices and daily records.
Ongoing now
- Status
- Operational
- Last checked
- Last outage
- Oct 10
All sources operational · 1 sources ·
Last 90 days
- Outage
- 2 entries
- Outage duration
- Duration unknown
- Maintenance
- 0 entries
- Days with outages
- Day 2
- Average outage duration
- Duration unknown
- Longest outage duration
- Duration unknown
Outages by weekday
Daily records
Aug 6 – Nov 3
Events
- ResolvedSome APAC users are seeing errors in ChatGPT Work, conversations, and GPTs.OpenAI
- ResolvedElevated errors with GPT-5.6 Instant in FedRAMP workspacesOpenAI
- ResolvedDegraded performance in Codex and Work mode: Dot turn failures, delayed responses, and reduced proactivityOpenAI
- ResolvedDot in Codex may fail to create new threadsOpenAI
Monthly records
About OpenAI records
These records collect incident and delay notices for OpenAI services such as ChatGPT and the API. New notices come from OpenAI’s official status page. Notices are grouped by outage, delay, maintenance and recovery wording in their titles and descriptions, with the original times shown. When no time is given, older notices show no time, while newly found notices show the first check time marked observed. Check the original notice for the precise impact on the services you use.
Source: OpenAI
About these records
Where do these records come from?
We check each company’s own official incident notices every 10–15 minutes.
What do outage, degraded and maintenance mean?
We use the words in the original notice: outage means an interruption, degraded means reduced service, and maintenance means planned work.
How is outage time counted?
We add time only when both the start and recovery times are known. Otherwise, we show duration unknown.
Which time zone is used?
All dates and times are shown in Japan Standard Time (JST).