Cursor and Together AI deliver real-time, low-latency inference at scale