Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Qwen 3.8 and Claude Opus 5 show why raw benchmark scores don't predict the bill (venturebeat.com)
6 points by ashurandi 36 days ago | hide | past | favorite | 1 comment


true, depends heavily on what kind of task it is. if you already have every part layed out what needs to happen tokens can be fairly low. but if for example want to build new feature or a new design, needs to do a lot of research (for me a lot design reference/mood boards )what is very token intensive




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: