Vendor record
OpenAI
Checked
Front door check counts and OpenAI’s own incident posts. Use them to investigate a dependency; neither confirms what happened in your app.
Front door checks
We send a request to a documented OpenAI endpoint and count the checks that got the answer it is documented to give.
Front door reachable in 2,873 of 2,873 checks with a clear answer, last 30 days, checked every 15 minutes from a single CheckUpstream location.
6 more checks gave no clear answer and are not counted.
- Endpoint checked
- api.openai.com
- Scope
- apiNo vendor region
- First check in this window
- Latest clear answer
Checks 15 minutes apart cannot see disruptions shorter than that. A check that reached the front door does not show that every OpenAI product, region or your app worked.
How we count front door checksShorter disruptions
Our checks are 15 minutes apart, so they miss disruptions shorter than that. Datadog runs a separate outage tracker that may show disruptions our checks cannot see.
Independent outage detection for OpenAI: Datadog’s Updog(opens in a new tab)CheckUpstream is not affiliated with Datadog.
OpenAI says
Incident posts from OpenAI’s status page, with its titles, impact levels and times. They say what OpenAI chose to post, not what reached your app.
A closed record shows when our record closed. Some older records closed when a post left the status page, without a resolution post from the vendor.
43
incidents OpenAI posted, last 30 days
Median recorded duration: 1h 30m. From 38 incidents with vendor-supplied starts and recorded closure times.
Most recent records · all dates
Widespread feature degradation
Not specifiedFirst seen by CheckUpstream
Closed in our record
Delayed Costs data in the Compliance API
Not specifiedStarted, per OpenAI
Closed in our record(1d 17h)
Limited availability on gpt-image-2.5-flare and gpt-image-2.5-sunburst
Not specifiedStarted, per OpenAI
Closed in our record(42m)
Degraded performance in Codex and ChatGPT Work
Not specifiedStarted, per OpenAI
Closed in our record(2h 55m)
Degraded performance in GPT-6.1 Sol Codex and Work mode
Not specifiedFirst seen by CheckUpstream
Closed in our record
15 older records
Workspace admins may be unable to manage plugins
Not specifiedStarted, per OpenAI
Closed in our record(1h 46m)
Some ChatGPT content may be unavailable on Android
Not specifiedStarted, per OpenAI
Closed in our record(3h 30m)
Delayed Costs data in the Compliance API
Not specifiedStarted, per OpenAI
Closed in our record(2d 22h)
Some APAC users are seeing errors in ChatGPT Work, conversations, and GPTs.
Not specifiedStarted, per OpenAI
Closed in our record(1h 43m)
Elevated errors with GPT-5.6 Instant in FedRAMP workspaces
Not specifiedStarted, per OpenAI
Closed in our record(28m)
Degraded performance in Codex and Work mode: Dot turn failures, delayed responses, and reduced proactivity
Not specifiedStarted, per OpenAI
Closed in our record(1h 8m)
Dot in Codex may fail to create new threads
Not specifiedStarted, per OpenAI
Closed in our record(1h 38m)
Unable to create new ChatGPT Work threads in latest version of desktop app
Not specifiedStarted, per OpenAI
Closed in our record(6h 28m)
Workspace Agents Degraded and Responses Impacted
Not specifiedStarted, per OpenAI
Closed in our record(1h 57m)
Computer Use Is Experiencing Issues
Not specifiedStarted, per OpenAI
Closed in our record(2h 54m)
Errors processing large PDFs in the Responses API
Not specifiedStarted, per OpenAI
Closed in our record(4h 34m)
Elevated error rates in ChatGPT Go conversations
Not specifiedStarted, per OpenAI
Closed in our record(1h 14m)
Elevated errors affecting multiple products including ChatGPT (conversations), Image generations, Spaces and Pages, API rehydration, and Dots.
Not specifiedStarted, per OpenAI
Closed in our record(38m)
Elevated errors in API Platform login and administration APIs.
Not specifiedStarted, per OpenAI
Closed in our record(1h 4m)
Elevated error rates in Codex Cloud
Not specifiedFirst seen by CheckUpstream
Closed in our record
Your code, your dependencies
See which of your code paths call OpenAI
Connect GitHub or GitLab to find where your public or private repository uses OpenAI. Review each match and your alert settings.
Free to start. No credit card.
Use this record on your site
Add OpenAI status to your page.
Choose the format your page supports, then copy the code. Badges show the recorded vendor status; they do not establish impact on your app.
Frequently Asked Questions
Is OpenAI currently down?
OpenAI's latest recorded status is Operational. Check the timestamp and linked source at the top of this page. A vendor status report does not establish whether your application was affected.
When did OpenAI last have an outage?
The most recent closed incident record began 1 day ago and spans 1d 17h. The vendor reported an incident. Closure may reflect a disappeared post or an expired tracking period; it does not confirm vendor or application recovery. Check the dated history and source links.
Does CheckUpstream publish an uptime percentage for OpenAI?
No. CheckUpstream checks OpenAI’s front door every 15 minutes and publishes how many checks reached it over the last 30 days, with the endpoint checked. Checks 15 minutes apart cannot see disruptions shorter than that. A percentage would claim more than the checks can show.
How do I monitor OpenAI in production?
Create your account, connect GitHub or GitLab and choose a public or private repository. Review the vendor matches and your alert settings. Available notification channels and history depend on your plan.
How does CheckUpstream track OpenAI?
This record separates vendor-reported incidents from independent checks. Sources and coverage vary by vendor: review the linked source, check time and measurement scope. Neither kind of evidence establishes SLA credit eligibility.