Embedding models are one of the key components of many LLM applications, including retrieval augmented generation (RAG).