Hi,
I have been using ChatGPT Plus for some time now, and from several tests I have conducted, I see that the maximum context window I can obtain is about 2k tokens.
Essentially, if I paste or attach any document longer than that, everything exceeding the window is not considered. I have noticed the same thing when creating GPTs; in this case, too, for the knowledge attached files, it can analyze and consider only a block of 2K tokens at a time.
For example, I had uploaded a file of all the posts on LinkedIn (over 200), and when asked how many there were (the posts were identifiable by an h1 format of date and title), it always answered 20, which means it could only count those within the 2k token window.
Is this just a problem for me? Have others experienced this?
The website states that the context window for ChatGPT Plus and Team is 32K.