3 ms·

For many tasks you don't really need big models. And relatively small model, quantized too can be run on your macbook (not to mention Mac studio).