Run GLM-4.5-Air(110B)on a 16GBRAM consumer machine

A researcher ran a 110B parameter AI model on a basic 2016 desktop with 16GB of RAM and a 6GB GPU by carefully controlling where data is stored — in GPU memory, RAM, or on the hard drive. Using four "placement laws" and a probe tool that identifies which model layers are most sensitive to compression, he achieved speeds that matched pre-registered predictions. Smarter data placement, not better hardware, is the key insight.