RAG specifics for GPTs - How many tokens retrieved?

When you’ve uploaded documents to a GPT and execute a prompt - is there any documentation on how much data will be retrieved via RAG? And what the excerpt size is?


hy, any update here?
m working on a similar item.
i have added a file with purpose assistants to assistant api with retrieval tool. And in msgs, for a given query i ask the assistant to get me 10-50 similar items but it return some where between 2-6