Curious to know how many people make use of OpenAI versus self deployed local models like say Gemma 2 or similar. I have a project that was using OpenAI but I hit some sort of request limit and I'm a bit jaded by that and my experiences with cloud costs over the years. I feel like Gemma 2 might serve my purpose well enough since I've already indexed the content and need it for RAG plus some summarisation. So curious to hear what others are doing.