A 64GB RTX 5095 GPU @ $2500 would pretty much collapse the cloud hosted AI market. Being able to run Qwen Flash Next/Qwen 4 Flash on my local workstation would eliminate the need to pay for cloud hosted AI for the average person/developer. But still API out periodically for architecture level decisions/layout.

Periodically I run into problems that I need Sol for, but I suspect Qwen Flash will meet most of my needs