After playing around with the Gemma 3n (4b) model on my Mac mini (M2 - 8/256) for a few mins, it seems like Ollama takes a sec for it to load, but after it’s “loaded the model” it runs pretty quickly… gonna use it through OpenWebUI rq..