3 ms·Cursor and Together AI deliver real-time, low-latency inference at scale2 points by inferhaven 2mo ago