Thunderobot Launches AI Master M7000 Laptop Capable of Local 120B Model Inference
Thunderobot has officially launched its AI Master M7000 mobile workstation in China, featuring AMD's Ryzen AI Max+ 395 processor and priced at 19,999 RMB (~$2,800 USD). Powered by Phison's aiDaptiv+ technology, the laptop can offload large language model weights to SSD storage to run AI models up to 120 billion parameters locally. Running 120B-parameter LLMs typically demands expensive enterprise GPUs with massive VRAM, but SSD offloading brings this capability into a portable workstation. This offers developers, researchers, and privacy-sensitive sectors an affordable way to run high-parameter AI locally without relying on cloud infrastructure. The workstation combines 64GB of RAM with a dedicated 768GB + 85GB AI SSD storage setup and comes pre-installed with the Herdsman local inference engine. Additional hardware features include a 16-inch 2.5K display, a 99Wh battery, an OCuLink expansion port, 2.5GbE LAN, and a overall weight of 2.45kg.
## BACKGROUND
Large language models require substantial memory bandwidth and capacity to hold model weights during execution. Phison's aiDaptiv+ technology addresses VRAM memory bottlenecks by integrating high-speed NVMe SSDs into the system's memory hierarchy, enabling cost-effective local inference for massive models.