Skip to content
IsAIDownNow

← All error guides

OpenAI API error 429: what it means and how to fix it

Quick answer: A 429 from the OpenAI API is not an outage. It means you're sending requests too fast, your prepaid credits have run out, or a spend or usage limit has been reached. The message tells you which. Rate limits clear when you slow down; billing and limit errors keep failing until you change your credits or settings.

The message you're seeing

  • Rate limit reached for requests
  • Slow down
  • Credit balance exhausted
  • Organization spend limit reached
  • Project spend limit reached
  • Organization usage limit reached

Is it down right now?

Live status, updated every minute. If it says “No problems”, the cause is most likely on your side: follow the steps below.

Which 429 is it?

OpenAI's error-code documentation lists several different 429 errors. Read the message (and the code, if your SDK shows it):

  • Rate limit reached for requests: you're sending requests too quickly. Slow down and follow the Retry-After header when it's there.
  • Slow down: your request rate went up too fast. Follow Retry-After, cut the rate, then increase it gradually.
  • Credit balance exhausted: your prepaid credits are used up. Add credits to continue.
  • Organization spend limit reached: your organisation hit the spend limit it set. Raise or remove the limit in organisation settings.
  • Project spend limit reached: the project hit its own spend limit. Raise or remove it in the project settings.
  • Organization usage limit reached: your organisation hit a usage limit assigned by OpenAI. Request a higher limit or contact support.

Fixing rate-limit 429s in code

  1. Respect Retry-AfterWhen the response includes a Retry-After header, wait at least that long before trying again.
  2. Back off exponentiallyWithout a header, retry after a short wait that doubles each time, with a little randomness added, and stop after a few attempts.
  3. Reduce concurrencyFewer requests in parallel, or a small queue in front of the API, keeps you under per-minute limits.
  4. Ramp up graduallyIf you're launching something that will send a lot of traffic, increase it step by step rather than all at once, which is what triggers “Slow down”.

Fixing billing and limit 429s

OpenAI's documentation is clear that retrying won't restore access for credit, spend-limit or usage-limit errors. Nothing will work until the credits or limits change. Check your billing and limits in the OpenAI platform settings for the organisation and project your API key belongs to.

Many older guides mention a message about having exceeded your current quota. OpenAI's current documentation lists the specific credit and spend-limit errors above instead, so search for the exact text you see.

Still getting errors?

If you're seeing 500 or 503 errors rather than 429s, that's a different problem on OpenAI's side: see our guide to OpenAI API 500 and 503 errors. And check the live status above to see whether the API is having a wider incident.

Questions

Is a 429 error an OpenAI outage?

No. A 429 is about your own requests, credits or limits. Outages usually show up as 500 or 503 errors and on OpenAI's status page.

Does a ChatGPT Plus subscription include API credits?

No. ChatGPT subscriptions and the OpenAI API are billed separately, so a paid ChatGPT plan doesn't add credits to your API account.

Why do I get “Slow down” when I'm under my rate limit?

OpenAI returns “Slow down” when your request rate rises too quickly, even below the limit. Increase traffic gradually and follow the Retry-After header.

Want to know when OpenAI API has a real outage? Follow our free outage alerts. We only post real outages, and again when they're fixed.

TelegramBluesky

Related error guides

Based on official documentation, checked 18 October 2026: OpenAI: API error codes. Error messages and limits change over time, so the provider's own docs are the final word.