6 ms·

I've recently learned and then observed that oMLX serves local models much, much faster than LM Studio.