alt
Hacker News
sourcecodeplz
•
today at 12:03 PM
•
0 replies
•
view on HN
still quite verbose at 140m output tokens, but this is on max thinking. high should do better.