Ziwen Xu
Region: US
Sunday 22 February 2026 08:38:42 GMT
6908
148
13
12
Music
Download
Comments
mark :
Quick question, have you had a chance to compare the effectiveness of a local vs online model? Only reason why I’m hesitant to host my own LLM is bc of more hallucinations, inaccuracies, and prompt injection risks. It saves a lot of money though so it could def maybe be used only for smaller or internal tasks (no internet communication to prevent risks)
2026-02-23 15:59:07
2
TRU :
nice, my guy — most on here are just larping or creating ads for shit they can’t do. would be good to add discussion on tok/s, quantization, vram / hardware impacts, etc. (just suggestions…i know you know)
2026-02-26 16:08:54
2
austin :
i thought gpt oss 120b was benched with one of the highest token per second speeds?
2026-04-13 15:46:46
1
Twe :
I just told codex to do it 😆. That is a nice set up user xp tho, monetize it my guy.
2026-02-22 21:30:00
1
To see more videos from user @ziwen.xu.00, please go to the Tikwm
homepage.
