Live status · OpenAI

Is OpenAI down?

This page asks OpenAI’s own status feed, straight from your browser, when you open it and once a minute while it stays open. It shows what OpenAI reports and never fills in a gap with a guess.

Not checked

Checking needs JavaScript, which is off or has not loaded. OpenAI’s own page has the answer: status.openai.com.

OpenAI’s status page

What this checks

The GPT models, the connection your AI tools use to reach them, and Codex. OpenAI’s feed does not give a status for each part. It lists open incidents and the parts each one affects, and this page keeps the ones that name “responses”, “chat completions”, “embeddings”, “realtime”, “batch”, “codex”, “conversations”, “login” or “voice mode”.

Open incidents

Anything open appears here after the live check.

Recent incidents

OpenAI’s feed only lists what is open right now, so this page cannot show past incidents. They are on status.openai.com.

If it is down

Each open incident below names the parts of OpenAI it affects. If none of them is the part you use, the problem is more likely on your side.
“Incorrect API key provided” is about the key, not an outage: see 401 Unauthorized.
If OpenAI reports no problems and it still fails, the cause is more likely on your side: your key, your code or your network. The error explainers cover the common messages.
A status page shows what the company has reported. A problem that has started but has not been reported yet will not show here, so if the answer above is no and something still fails for everyone you know, check again in a few minutes.
Do not rewrite working code to chase an outage. Commit what you have now, so that when the service is back you can tell whether anything you changed in the meantime broke something.
Down? Save your place before you switch tools: Cold Handoff writes the note you paste into the next session, so it picks up where this one stopped instead of starting over.

Where this comes from