45 TOPS of computing power, runs 9B models natively, supports up to 32B quantized large models; up to 64GB memory + 2TB storage. Designed for individuals / small teams, focusing on daily office work, document analysis, content creation, and lightweight local model debugging.