Is anyone here into local LLM tech?
I saw
@NVK posted some good stuff about “ds4” that I’m going to test out.
I’ve been running a few smaller models (smaller than deep seek 4) for some personal projects:
- tts for tweets -> mp3
- automating canceling/resuming Netflix
- a self-sovereign rocket money clone
For me these all work great with gemma4 & qwen3.6 models. Fast inference, accurate results.
I’m wondering if DS4 would be even better, or could do all of the above plus more.