2 ms·It runs locally, so there's no need for them to spend money on a server-side inference layer.by simonw 2y agoIt runs locally, so there's no need for them to spend money on a server-side inference layer.