can you add deepseek q8 ?

[jonipupet]

for max studio m3 ultra 512gb q8 is best choice which is around 300gb
still much root for qually can make it 400-450gb q12 ))

[antirez]

It is a base model, not fine tuned, according to the model card.

[antirez]

Mmmm checking the model card I'm no longer sure if they mean "base" as "full weights precision" or base as what traditionally this used to mean. But looking at the kind of benchmarks of the base model, I believe it is NOT fine tuned. So not worth supporting here.