Its one thing to force a new model on users. But then tell them to use 3.5 while that new model is on cooldown. Started a new conversation, since switching to 3.5 is not working in the current conversation (it wont let you enter the text). After 5 posts in the new conversation in 3.5 it too would no longer allow furtherance of the conversation. (again not letting the text be entered). I was happy with 3.5 exclusively, but now ChatGPT is just overall completely unusable for anything.
Same. It’s utterly useless right now. I’m sure it’s a way to force more updated subscriptions, but I tried that a month or so ago and it was also similarly limited. Useless.
I’ve since installed a local GPT using Ooba at github – oobabooga/text-generation-webui) While it’s still not as polished as GPT35, it good enough for now. We’ll see how long OpenAI takes to figure this out.