I am trying to run a chat completion via PowerAutomate (I used both the OpenAI connector and an HTTP POST) but both fail due to timeout limits, when using the gpt-4 model. The flow runs a number of retries until it eventually fails.
gpt-3.5-turbo works mostly and text-davinci-003 work all the time. However, gpt-3 cannot complete the prompt to my satisfaction, therefore I need to use gpt-4.
I have also run the same request via python, using gpt-4 and that worked.
Do you have any ideas what I could do?