Just be careful following some of the advice in these videos. In my experience there is “no one size fits all” when it comes to local models. It’s far more nuanced and there are many factors playing into it. With a Mac it’s easier to predict how local models will perform due to the operating system, CPU and platform being known quantities with identifiable variations but that’s not the case for Windows machines, not even where the hardware appears to be the same.
I still think local models - either running directly on the hardware you have now, or on emerging dedicated appliances - is the direction for agentic coding.