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

Thank you. Which is currently the most capable version running reasonably fast on a 3090 (24GB of VRAM)?


The Llama distilled version Q4_K_M should be reasonably fast and good!!




Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

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

Search: