OpenAI API outage — what to do when the API is down

You're getting 503s, 529s, or complete timeouts from the OpenAI API. Check the [live OpenAI status](/status) on this page — it updates every 60 seconds from OpenAI's official feed and shows exactly which API components a

OpenAI API outage — what to do when the API is down

You're getting 503s, 529s, or complete timeouts from the OpenAI API. Check the live OpenAI status on this page — it updates every 60 seconds from OpenAI's official feed and shows exactly which API components are degraded or down. If you see red indicators, it's an outage. If everything shows green, the problem's likely on your end.

How to check if OpenAI is actually down

Start with our live OpenAI status monitor at the top of this page. It pulls directly from OpenAI's infrastructure status and updates automatically. Look at the "API" row specifically — green means operational, yellow means degraded performance, red means outage. You'll see timestamps showing when any incident started.

Cross-reference with status.openai.com. OpenAI posts official incident reports there with specific details: which models are affected, which regions, estimated recovery time. During major outages, you'll see entries like "GPT-4 API elevated error rates" or "Completions API partial outage — US East." These reports update every 15-30 minutes during active incidents.

Check Twitter/X by searching "OpenAI API down" for real-time reports from other developers. If hundreds of people are posting error screenshots in the last 10 minutes, it's a widespread outage. If you see nothing recent, dig deeper into your own setup.

Is it OpenAI or is it you

Before assuming it's an outage, isolate your environment:

  • Test with curl directly — strip out your application code and hit the API with a basic curl command: `curl https://api.openai.com/v1/chat/completions -H "Authorization: Bearer YOUR_KEY" -H "Content-Type: application/json" -d '{"model":"gpt-3.5-turbo","messages":[{"role":"user","content":"test"}]}'`. If this fails with the same error, it's not your code.
  • Check your API key status — log into platform.openai.com/api-keys and verify your key hasn't been revoked. Check Usage → Limits to confirm you haven't hit rate limits or exhausted credits.
  • Review recent error codes — platform.openai.com/usage shows your error rates. If you're seeing 429 errors specifically, you're hitting rate limits (see OpenAI API error 429 and other API errors). Error 503 or 529 during an outage is different from 429 rate limits.
  • Test different models — try gpt-3.5-turbo instead of gpt-4. Sometimes specific models go down while others stay operational.
  • Check your network — corporate VPNs, firewalls, or cloud provider network issues can mimic API outages. Test from a different network or machine.

What to do during an outage

When you've confirmed it's an actual OpenAI outage, implement these strategies:

  • Implement exponential backoff immediately — retry failed requests with increasing delays: 1 second, 2 seconds, 4 seconds, 8 seconds, up to 60 seconds max. Don't hammer the API with constant retries every 100ms. That makes outages worse for everyone.
  • Add request queuing — buffer incoming requests in a queue instead of rejecting users. When the API recovers, process the queue gradually to avoid triggering rate limits.
  • Set retry budgets — limit total retry attempts to 5-7 per request. After that, fail gracefully with a clear user message: "OpenAI's API is experiencing issues. Your request will be processed when service resumes."
  • Cache aggressively during degraded performance — if you're seeing 70% success rates, cache successful responses more heavily and reuse them where acceptable.
  • Monitor OpenAI's status page — refresh status.openai.com every 5-10 minutes for incident updates and estimated recovery times.

Don't refresh your API keys, recreate your account, or open multiple support tickets. These actions waste time and won't fix an infrastructure outage. For actual support issues unrelated to outages, see How to contact OpenAI support.

Questions people actually ask

Q: Is the OpenAI API down right now?

A: Check the live OpenAI status at the top of this page. It shows real-time API status pulled every 60 seconds from OpenAI's official feed. Green means operational, red means outage.

Q: How long do OpenAI API outages typically last?

A: Minor incidents resolve in 15-45 minutes. Major outages can last 2-4 hours. Check status.openai.com for specific incident timelines and recovery estimates.

Q: Will OpenAI refund API credits for downtime?

A: OpenAI doesn't automatically refund for outages. For extended downtime affecting your production systems, contact billing through platform.openai.com, but don't expect guaranteed refunds.

Q: Should I switch to a different model during an outage?

A: If GPT-4 is down but GPT-3.5-turbo shows operational, yes — temporarily switch models. Check which specific models are affected on status.openai.com.

What to remember

  • Check the live status monitor on this page first — it updates every 60 seconds
  • Use exponential backoff and retry budgets, not constant hammering
  • Queue requests during outages instead of rejecting users
  • Test with curl and different models to isolate the problem
  • Don't refresh API keys or recreate accounts during outages

---

*openai-support.com is an independent resource, not affiliated with OpenAI Inc.*

Related help