The are the way to go, but hardware pricing makes it cost prohibitive to match frontier capabilities, which makes decomp and recomp difficult. Or rather, require more skill and take longer, introducing some barriers. We all know the eye watering prices GPUs and RAM cost now, to say nothing of UnifiedMenory devices like the Strix. Most people just don’t have a spare $10K to throw at an inference box that can run (say) a 300B model. Coming at this from another angle tho, I note that according to Steam stats, the most popular GPUs are in the 8Gb and 16Gb range…which is smack bang in the VRAM range of what a good MoE (say, Qwen 3.6 35B-A3B) runs on (for 8gb) or arguably the best local dense coder (Qwen 3.8 27B) can live at. I think there’s probably room for great LLM improvement in the 8-16gb range…and given that most consumer pcs and laptops live there, I’m sanguine the open weight devs might make it happen. But I also suspect the gate keeping of high end capacities is intentional. We’re probably note going to have the ability to vibe code Mario into Elden Ring again…not without significantly more skill. Dunno.