Comment by Svoka

You can run smaller models on MacbookPro with ollama with those speeds. Even with several 3k GPUs it won't come close to 4o level.