Status code 503: That model is currently overloaded with other requests

I get this error when making an API call to Davinci model 3. Given that the GPT completion is part of the front end user experience for my site, this is a critical error as it disables the site completely. What’s going on?

Status code 503
  "error": {
    "message": "That model is currently overloaded with other requests. You can retry your request, or contact us through our help center at if the error persists. (Please include the request ID 7ef84fe5909aa6768fb8d39a081a5ccc in your message.)",
    "type": "server_error",
    "param": null,
    "code": null

If OpenAI can’t maintain the capacity for GPT-3 requests they really need to stop taking on new accounts until they can keep up or take capacity off of ChatGPT. It’s not fair that paid users don’t get a reliable service - there was another outage just 2 days ago.


I just came here to make a post about this, I’ll go ahead and tag along to this one.

I agree that it is unfair, we just scaled to $2,000/week in API calls and would really appreciate if our users didn’t have these outages.

Inside of our Facebook group we actually started getting a ton of posts over the last couple days about documents not being generated. I knew this was a problem sometimes, but no one ever posted about it that’s how rare it was.

I decide to take a look for myself and see the same 503 error code. And when it’s not the 503 error code, it’s a different one that I’m making too many request at once, which there’s no way is true.

Feeling… Frustrated, but not too upset because things are early. They do need to chill out on taking on more users though until they scale up their servers


Did you use an app to make use of the gpt? like github apps?

Hey @pencilllc, welcome to the OpenAI community.

And absolutely, you can utilize GPT to build some really powerful apps.

In terms of what I used to build mine, I use the React programming language, firebase for authentication, OpenAI for my AI of course and express.js for by backend. Highly recommend this stack if you are a beginner.

That’s pretty cool man. What application did you built? Love to try it out

Seeing this 503 error as well today. Any way to mitigate what is going on here?

Nope, but apparently it’s been fixed (

This is really, really poor. How are we supposed to integrate GPT3 if half the time it’s not even available?

AGAIN? It’s a joke at this point.

All of you that pay OpenAI for the use of any API should demand the company to protect your investment in the technology against those that dabble / play for free…


All of you that pay OpenAI for the use of any API should demand the company to protect your investment in the technology against those that dabble / play for free…

Another outage! Keep getting 503 errors and 429 Errors on the GPT-3 API which given that I’m in testing cannot possibly be correct. Anybody know anything about openai addressing these issues?

P.S. the status page indicates all systems are operational…

Same, I’m getting a lot of 503 errors and 429 Errors for completions GPT 3 endpoint.

429’s and 503 are becoming way too common. We really need a dedicated, robust server for paying customers only! Customer support never replies. I reply to them, and their next email is “Are you satisfied with our help? Thank you!”

I really hope they aren’t using AI to reply to us, ironic, but extremely unhelpful.


Yeah this is concerning.
Are we able to

  • make a fork of their GPT platform/source and
  • deploy it to our own servers on say AWS so we
  • have our own instance serving our own requests
  • for only our app
  • on a completely separate server?

Similar to how Wordpress lets users deploy the entire Wordpress platform/source to their own servers?

No, and that will never be possible

Why do you think it will never be possible? :thinking:

its still here, having these issues every single evening

1 Like

I’m having a ton of these issues myself. iIs pretty bad since my app makes 3 requests to accomplish a task so it has to pass all 3.

Time to start using those investment dollars guys.

1 Like

Finally converted to a paid account but getting 503s today :pensive:. It’s going to be hard to build anything on top of an unreliable API.