Based on your evaluation, it sounds like it will run inference at speed similar to an M4 Max and also allow "startups" to experiment with fine tuning larger mod

Based on your evaluation, it sounds like it will run inference at speed similar to an M4 Max and also allow "startups" to experiment with fine tuning larger models or larger context windows. It's the best "dev board" setup I've seen so far. It might be part of their larger commercial plan but it definitely hits the sweet spot for the home enthusiast who have been pleading for more VRAM.