Not trying to intrude, especially not knowing FSP policies or stance on the subject. Yet, I’m curious.

1. What HW is used to run FSP AI? Mac Studio? DGX Sparc array? Else?
2. What base models are used? The names, quantizations? What are the short to mid term plans of upgrading the base models as well as HW?
3. Any plans to let FPS users copy existing, past or future models?

Not sure how damaging to FSP revealing the above information can be, so say what you can, if at all. My main motivation is to see if multiplying FSP tech is helpful to the cause and when.

Thank you!