Although both RAG & long-context LLMs aim to solve the problem of answering queries over a large corpus of text, a direct comparison on a common task is still lacking, making evaluation challenging.