Anyone experienced GPT o1 pro starting to glich and completely missunderstand even simple instructions? This happens when I work on long texts and long conversations. Looks like it is something with context window. But sometimes even I promt to forget all instructions or try to reset it in the chat and asking making summaries or refferences so that I can keep working in same conversation and keep GPT understand needed context and still have space in context window for future work, then after giving some simple task, it gliches and perfoms way another tasks similar what was mentioned long time in the conversation history.
It’s anoying, because I have to start new chat and teach GPT to new data, since it complicated matter.
I understand the context window limitations, but there are something more on it, like there are some bug or glich when context is filled and GPT needs to do summarise in the background, somehow it makes huge mistakes there.