> For my local setup, I’m currently [..] and LM Studio as the inference server, ... | Hacker News

Hacker Timesnew | past | comments | ask | show | jobs | submit

andix 23 days ago | parent | context | favorite | on: Running local models is good now

> For my local setup, I’m currently [..] and LM Studio as the inference server, although it would likely be faster if I just used llama.cpp directly

Is there any truth to this claim? LM Studio uses llama.cpp to run the models. I guess the overhead of LM Studio should be minimal.

After all LM Studio is a really easy way to host models, are there really major drawbacks?

cuvinny 23 days ago [–]

LM Studio has a lot less tuning options when you launch it. Also it is precompiled so you don't have the latest (and sometimes buggy) releases so you may have to wait a few days to try out a MTP or a new model. LM Studio is easier though.

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