(mistral.ai)

701 points mfiguiere | 1 comments | 21 May 25 14:21 UTC | HN request time: 0s | source

Show context

ManlyBread ◴[21 May 25 18:07 UTC] No.44054351[source]▶

>Devstral is light enough to run on a single RTX 4090 or a Mac with 32GB RAM, making it an ideal choice for local deployment and on-device use

This is still too much, a single 4090 costs $3k

orbisvicis ◴[21 May 25 18:37 UTC] No.44054679[source]▶

Is there an equivalence between gpu vram and mac ram?

replies(1): >>44056661 #

1. viraptor ◴[21 May 25 21:52 UTC] No.44056661[source]▶

For loading models, it's exactly the same. Mac ram is fully (more or less) shared between CPU/GPU.