It looks like in a few years, the basic large language models (LLMs) we use will get commoditised, and it won't really matter which one you pick. The next big thing could be LLMs that use Retrieval-Augmented Generation (RAG), which means they need …