15

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?

16

No, and that will never be possible

18

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

19

its still here, having these issues every single evening

1 Like
20

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
21

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.

1 Like
22

@manphilip nice seeing you here!

23

They just announced they’re making an API for ChatGPT. Why not make sure that GPT3 actually works reliably before creating something new?

24

I am getting 503 requests half the time with the API. Very unreliable feature from a business standpoint.

25

Spent a lot of time on my project thinking of shelving it because the api is still unreliable :frowning: Ironic how Small Altman was expresident of YC and does not seem to care about his customers lol

26

the API is more reliable if the prompts are longer and more descriptive, at least at this moment

27

thanks thats useful to know

28

I am also experiencing this:
“That model is currently overloaded with other requests. You can retry your request, or contact us through our help center at help.openai.com if the error persists. (Please include the request ID XXXXX in your message.)”
This happens every 10th request at the moment, on a paid account. Not sure how I am supposed to be using this for anything that isn’t a fun side project.
Is there anything I can do?

29

I have a wide spectrum of problems with the playground and the model. This is independent of the browser and internet parameters. For three days, if there are no results after submit, then there are also various error messages such as delays, overloads or other messages. Tech support doesn’t respond at all to all this and pretends it’s fine.

What is more, my post is tagged as spam!

https://community.openai.com/t/text-davinci-003-and-3-days-of-appocalipse-different-problems-and-indifference-of-technical-assistance/58622/1

30

Oh wow, again? OpenAI needs to get a grip.

31

Nope, My prompt size is relatively large but still happening same.

32

They are shooting themselves in a leg by branding the product and not being able to serve… Same happens with OpenAI API and Azure API which is designed for businesses. It’s a pity, really.

33

what kind of work are you using chat gpt if you dont mind my asking?

34

To check the current status of the OpenAI API, head to: https://status.openai.com/

1 Like
35