With the release of some new platforms having 128GB or more of unified memories, I am wondering if I should spend some money on those.

I am looking for feedback and stories of people who bought hardware for local-first AI.

- Did you identify a sweet spot in model size, and therefore hardware sizing?

- Did you continue to use cloud / frontier models?


Comments URL: https://news.ycombinator.com/item?id=50578897

Points: 2

# Comments: 1