Tablet LLMs: Running AI just as powerful as my PC

The high-performance workstation scene in my office usually features three formidable Strix Halo machines. Until recently, however, one device on the team seemed to be operating at a fraction of its potential: the Asus ROG Flow Z13. It was certainly taking up space, but it was doing the absolute least amount of demanding work—primarily playing games like Death Stranding 2 and nothing else.

This setup, while powerful, wasn’t pushing its limits in the way I was currently interested in exploring. The real test came when I decided to shift focus from pure gaming benchmarks to the burgeoning world of Artificial Intelligence and Large Language Models (LLMs).

The next step involved setting up a more rigorous environment for experimentation. I began installing Lemonade, a tool designed to put demanding computational load on systems while running LLM processes. This wasn’t just a casual experiment; it was an attempt to determine how well the mobile powerhouse could handle the heavy lifting required by modern AI applications.

With Lemonade installed across my other testing rigs, the focus turned entirely to the ROG Flow Z13. I wanted to see if this sleek, portable tablet possessed the necessary horsepower and thermal management to compete effectively in the AI landscape. Could it transition from a gaming rig into a serious LLM testing platform?

The goal was simple: measure the tablet’s true capabilities when faced with the intensive demands of running complex AI software locally. It was an exercise in checking whether cutting-edge mobile hardware could keep pace with desktop-class performance when the workload shifts to computational intelligence.

This experiment aimed to uncover the potential hidden within the Flow Z13, pushing it beyond its usual gaming role and revealing its capacity for serious computation. The results of this benchmark will tell us a lot about the future of local AI processing on portable hardware.

By Liam Brooks