6 ms·

Give a task you have to 3 different models and see what actually works for you. There are no good benchmarks.