Live status
Is the OpenAI API down right now?
The OpenAI API: no problems in our tests in the last 15 minutes · official status (status.openai.com): operational · checked
Updated . While people are reading it, this page is refreshed about once a minute.
Our tests of OpenAI's models
We send "Reply with OK" to GPT-6 Luna every 5 minutes and GPT-6 Astra every 30 minutes, and time how long each takes to start answering and to finish. The verdict covers the last 15 minutes: our last three tests.
Last test
- Answered, last 15 min
- 100%
- Answered, last hour
- 100%
- Answered, last 24 h
- 100%
- Median first token, 24 h
- 0.8 s
The last 24 hours
| Model | Last 24 h | First token | Whole reply | Last test |
|---|---|---|---|---|
| GPT-6 LunaEvery 5 minutes | 43 of 43 answered | 0.8 s | 0.9 s | AnsweredThrough OpenRouter, |
| GPT-6 AstraEvery 30 minutes | 7 of 7 answered | 1.3 s | 1.4 s | AnsweredThrough OpenRouter, |
No failed tests in the last 24 hours.
What OpenAI's status page says
OpenAI's own status page, read at most once a minute. It reports incidents once OpenAI's team confirms them, so it can lag behind what users and our tests see.
From status.openai.com, read
- Chat CompletionsOperational
- ResponsesOperational
- Fine-tuningOperational
- EmbeddingsOperational
- ImagesOperational
- BatchOperational
- AudioOperational
- ModerationsOperational
And 3 more parts, all operational.
Open incidents
None listed right now.
Resolved on status.openai.com in the last 14 days
Issues with Codex Started Sep 25, 22:58 UTC · resolved Sep 25, 23:54 UTC
Elevated Error Rates on GPT-6 Astra ProStarted Sep 24, 18:59 UTC · resolved Sep 24, 19:46 UTC
Mobile users unable to see Work Mode and the Model PickerStarted Sep 23, 23:25 UTC · resolved Sep 23, 23:49 UTC
Elevated Error Rates for ChatGPT across Plus and Pro plans.Started Sep 23, 09:13 UTC · resolved Sep 23, 09:54 UTC
Elevated Error Rates for ChatGPT Work, across Plus, Pro, Business, Enterprise and Education plans.Started Sep 22, 15:04 UTC · resolved Sep 22, 15:58 UTC
What our test covers, and what it doesn't
This page is about OpenAI's API, which is what our test calls: the same kind of request your code sends. ChatGPT is a separate product on top of the same models, with its own page here and its own parts on OpenAI's status page.
Each test is one message of a few words, with the reply capped at 32 tokens, tried once, with no retry and no switch to another route. A test counts as failed when the model refuses it as busy (a rate limit or "overloaded"), breaks (a server error), drops the connection, or takes more than 30 seconds. A test our own request broke doesn't count against the OpenAI API.
What to do now
Check the two answers above. If both look fine, read the error your code got: its code says whose problem it is.
Retry with a growing wait between tries, and cap the number of tries.
If one model fails, try another GPT model for now.
For work that can't wait, ask another company's model.
Ask another model
llmwise has Claude, Gemini, Grok and more in one chat, so an outage at one company doesn't stop your work. The free trial is 5 messages, with a sign-in link by email and no card.
The OpenAI API errors, and how to fix them
The errors people see most with the OpenAI API, what each one usually means, and what to try.
429 "Rate limit reached"
Your organization hit its requests or tokens per minute.
Retry with a growing wait between tries, and send fewer requests or tokens a minute. Your limits are in the platform's limits page.
429 "You exceeded your current quota"
Your account's quota is used up: a billing problem, not an outage.
Check your billing and spending limit. Retrying won't help.
500 or 503 "The server had an error" or "overloaded"
Something failed on OpenAI's side, or it's too busy.
Retry after a short wait. If many requests fail, check the status above.
401 Unauthorized
The request's key was refused: a setup problem, not an outage.
Check that the key is current and belongs to the right project and organization.
Timeouts on long requests
A long answer took longer than your client waits.
Stream the response, raise your client's timeout, or ask for less output.
Other AI status pages
Questions
Is the OpenAI API down, or is it my code?
If our test passes and status.openai.com shows Chat Completions and Responses working, look at your own requests: the error's code and message, your usage limits, and your organization's settings.
What does an OpenAI 429 error mean?
It depends on the message. "Rate limit reached" means too many requests or tokens a minute: slow down and retry. "You exceeded your current quota" means your account's spending limit or balance: retrying won't help until you change it.
Why does this page show different parts from the ChatGPT page?
OpenAI's status page lists API parts (Chat Completions, Responses, Batch and more) separately from ChatGPT's. We show the API group here, and ChatGPT's on the ChatGPT page.
How does your test call OpenAI?
We send one tiny request to GPT-6 Luna every 5 minutes and GPT-6 Astra every 30 minutes, the way llmwise's chat sends them: through OpenRouter or directly to OpenAI. Each result says which route it took.
Claude, GPT, Gemini, DeepSeek, Grok, Kimi, and GLM, in one chat.
See what a message costs before you send it. Free is 5 messages to try; sign in with an email link, no password or card.