Local embedding model
Also known as: local embedder
A local embedding model is an embedding model you run on your own machine, so your semantic search and RAG don’t depend on a cloud API.
- Runs via Ollama, llama.cpp, or a small framework.
- Produces the same kind of vectors as a cloud embedder.
- Keeps your text private and offline-capable.
The RAG foundation: for local RAG, pick a solid local embedding model first — it's the quiet component that decides whether retrieval is any good.