Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

So far, I like the answer quality and its voice (a bit less obsequious than either ChatGPT or DeepSeek, more direct), but it seems to badly mangle the format of its answers more often than I've seen with SOTA models (I'd include DeepSeek in that category, or close enough).


Which host did you use? I noticed the same using parasail. Switching to novita and temp 0.4 solved it.


The host was Moonshot AI at Kimi dot com :)




Consider applying for YC's Winter 2026 batch! Applications are open till Nov 10

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

Search: