Hacker News Reader

Comment by ant6n | Hacker News Reader

Just get a Mac Studio with 512GB RAM and run a local model when the internet is down.

Which local model would you recommend that comes close to cursor in response quality? I have tried deepseek, mistral, and few others. None comes close to quality of cursor. I keep coming back to it.

Possibly useful comment on local models, perhaps also fitting on machines with less ram:

https://news.ycombinator.com/item?id=43340989

Back up your $20 a month subscription with a $2000 Mac Studio for those days your internet is down.

Peak HN.

Lol he suggested a $10k Mac Studio

But you can at least resell that $10k Mac Studio, theoretically.

Trying to do that with an M1 laptop of 32 GB and it's hard to get even 1000 euro's for it in the Netherlands whereas the refurbished price is at double of that.

Even more absurd is that Mac Studio with 512GB RAM costs around $9.5K

> Peak HN.

But, alas, not a single upvote.

2000$? You wish!

Lol, not sure where I got the 2k from. Brain fart, but I'll let it stand :D

Maybe this "backup" solution.. developed into commodity hardware as an affordable open source solution that keeps the model and code locally and private at all times is the actual solution we need.

Lets say a cluster of raspberry pi's / low powered devices producing results as good as claude 3.7-sonnet. Would it be completely infeasible to create a custom model that is trained on your own code base and might not be a fully fledged LLM but provides similar features to cursor?

Have we all gone bonkers sending our code to third parties? The code is the thing you want to keep secret unless your working on an open source project.

Can one run cursor with local LLMs only?

... to make *completely* sure that they forgot how to program?

But then I’d be using a Mac, and that would slow my development down and be generally miserable.

lol

Reply on news.ycombinator.com