I’m aware that LLMs have context size limits, and I know the usual workaround is to open a fresh chat when a conversation grows too large. That part makes sense.
What doesn’t make sense is the interface collapsing long before any model limit is reached.
When a chat builds up a long message history, the UI becomes progressively unstable. In my case, it goes beyond minor slowdown — it turns into a full usability failure:
-
Text input delays or locks up while typing
-
Deleting or editing a prompt can take several seconds per keypress
-
Highlighting messages or copying text often doesn’t work at all
-
The whole interface sometimes becomes temporarily unresponsive
-
And recently for the first time: a chat froze entirely and never returned to normal, leaving my last prompt unanswered indefinitely
I’ve tested multiple other AI chat platforms with very long conversation histories, and while some show slower model responses, none grind the UI to a halt the way this does. No freezing, no input lag, no text selection breaking.
To illustrate the behavior, I uploaded a screen recording here:
youtu.be/p2uZ7TuVUro
I’m less concerned about how much history the model can ingest, and more about why the frontend becomes unusable during normal interactions — even for paying Plus users.
Curious to hear if others have experienced similar UI-level instability in extended chats.