Try https://github.com/SecureAI-Tools/SecureAI-Tools -- it's an open-source application layer for Retrieval-Augmented Generation (RAG). It allows you to use any LLM -- you can use OpenAI APIs, or run models locally with Ollama.