I’ve been working with the API and have a functioning application that works on GPT-3.5 and GPT-4 (current 0314 model). I have been using 3.5 because it is much faster than 4 has been, but we want to upgrade to 4 as it formats the responses more reliably for our needs. When I tried using gpt-4-0613 today, I am finding that it starts responding, then within 2 seconds or so it just stops. No error, I just get the done message from the stream.
Anyone else experiencing this, or know what to troubleshoot as to why it’s just giving up?