← sources

Which model/setup do you recommend for RTX5070 (non TI)

source: reddit (reddit:LocalLLaMA)published 2026-10-28 02:04ingested 2026-10-29 02:42

retained text

Hi,

I'm a bit of a noob when it comes to running local LLMs, I have tried a few things, but everything was miserably slow on my setup. I have an RTX5070 with 64g system memory.

Recently I have experimented with unsloth, tried a few model messed with settings, bit it only gotten worse. I feel like I can achieve something better than half an hour to an hour response time for anything non trivial.

Could you help me with recommendation on which model/setup should I try to achieve a usable speed? 

Or if you could direct me where to even try to look I would be grateful.

(I tried to search the sub, but all I found was related to 5070 TI

document fetched 2026-10-29 00:44 · canonical https://reddit.com/r/LocalLLaMA/comments/1xhbruy/which_modelsetup_do_you_recommend_for_rtx5070_non