Gemini 1.5 Pro/Flash crashes with books or text inputs over 100k tokens

I wanted to make a summary and ask some questions about a book (The DIary of Anne Frank) but every time I send it, the model starts processing the answer but stops at 3-4 seconds.

Why? The book is 170k tokens which is far from the 1 mln tokens limit. I’ve also tried deactivating all the safety filters or sending the book via text form in the prompt but it still leads to the same exact issue.

Hi @Magrolino,

Welcome you to the community!

I understand you were having some trouble summarizing and questioning the book. I went ahead and tried to replicate the issue myself, and I’m glad to report that I was able to successfully summarize and ask questions about it. Here are some screenshots to show you.

Let me know if you have any other questions!


2 Likes

Thank you very much for the answer, Siva!

Encouraged by your screenshots, I tried again and faced the same issue but I have found a fix. If I send another message to the model after the freeze, the model will be able to process the input and actually reply in its second attempt.

Thank you!

1 Like