No history anymore, constant network errors after 80% of generating the answer.
More info : Chat Gpt 4 gives ‘Network Error’ after 80% of generation
Status (https://status.openai.com/) is green but there is still no history and crashes while answering.
+1 and i want to add that it really feels like a bait and switch to limit the plus subscribers to only 25 questions with GPT-4 per 3/hr block. I barely even use it because Ive had too many times where I’m just getting to some useful material and i get locked out. Totally breaks my flow, gotta stop to manually summarize and recontextualize for GPT3.5, break tasks into smaller chunks, etc. And then i have to waste half my allotted messages when I get more just to feed the updated information into GPT-4 again.
Two requests that would help a lot would be in-chat model switching and a counter until lockout. If I could give most of my “change this in specific way” or “construct outline in format B” type instructions to 3.5 and reserve 4 for the main questions within one historical context, that’d be super sick.
The lack of history seems like an unintentional technical issue rather than a choice, so I cant hold that against them (as a dev, ive accidentally broken things for customers before, and i wouldve wanted the same courtesy). I noticed that they’re not mentioning it on the status portal too though, and that worries me. The service is degraded, has been for days, and that should be acknowledged.
Edit: History is and has been back for a little while now, and at least for myself I haven’t noticed anything missing.
This whole thing really illustrates the need for a built in export and export all feature though, as markdown and json. There are of course ways to do it with the console, but expecting the average user to do that is a bit much (and whether it’s allowed/supported is a grey area).
Now I’m gettings constant : “We’re currently processing too many requests. Please try again later.” which is why I paid for PLUS … to no avail.
Status continues green.