obsessed with it, 128gb gpu+cpu, decent mem bw, runs most of the llm models locally on roo code… useful token/sec on qwen3.6, and it looks the coolest, asus rog flow z13-kjp