Post #4257885

2026-05-02 14:10 UTC

@chrislw@podcastindex.social My recent experience with Qwen3.6 35B running locally on a DGX Spark tells me that the local models are virtually on par with the hyperscale models for many tasks like coding. A developer with a 128 gig Mac mini/studio/MacBook can be pretty much self sufficient and not pay those guys a dime. I think that’s what we are seeing the beginning of now. You can’t find a big RAM Mac anywhere right now. All sold out.

Replies (2)

@chrislw@podcastindex.social 2026-05-02 14:48

@dave@podcastindex.social yeah, I will probably be setting up a NAS in the future. Might go with a higher end UGreen enclosure, upgrade the RAM, and then get several enterprise level 20TB drives. Not sure if I will flash a new OS like Unraid on it.

Open ##4257886
@russell@podcastindex.social 2026-05-03 04:48

@dave@podcastindex.social @chrislw@podcastindex.social this issue isn’t can you run a LLM on your machine, it’s can the LLM remember conversation and code from 6 months ago. This is where neural network databases come in handy, but today we don’t really have that. What we have is databases that can store history but you have to have your reasoner script tell and supply the LLM with the information each time. It’s like a pre-think instead of the LLM looking automatically at the stored data overtime. It’s really difficult atm.

Open ##4257888