Both can be true at the same time. I currently wouldn't waste my time with open models for almost all use cases, but they're crucial from a data privacy and competitive perspective, and I can't wait for them to catch up enough to be as useful as the current frontier models.
I've found qwen3 to be very usable on my local machine (a Framework Desktop with 128gb RAM). I doubt it could handle the complex tasks I throw at Claude Opus at work, but it's more than capable of doing a surprising number of tasks, with good performance.