RAG: More documents may reduce reliance on larger LLMs
A CMU researchers’ paper, dated January 14, 2026, evaluates expanding the retrieval corpus as an alternative to scaling up the model. According to the post, a medium-sized model with more documents may match a larger model with a smaller corpus.
Source: Menos LLM, mais documentos: uma abordagem de scaling para RAG (arxiv.org). Text prepared with AI from this source.
What happened and what to do
The paper “Less LLM, More Documents: Searching for Improved RAG” by CMU researchers, dated January 14, 2026, studies whether expanding the retrieval corpus can be an alternative to scaling up the language model. According to the post about the work, a medium-sized model with a large corpus may match a larger model with a smaller corpus.
A company can test this approach on its own use case: organize and expand relevant materials, instrument ingestion and search, and compare models and corpus sizes using representative questions. Quality, latency, and cost metrics can help determine whether to invest in more documents, a larger model, or improvements to retrieval.
How the consultancy can help
Wendelmaques can diagnose your RAG use case, define an evaluation using representative data and questions, and propose an implementation for ingestion, retrieval, and monitoring. The scope can include operating the solution on your infrastructure.
Next step
Send a short description of your search or RAG use case to receive a scoped proposal for diagnosis and implementation.
Let's assess your project
I reply with scope and a proposal.