
The Model That Would Not Speak: System One AI on Your Own Machine
One support ticket, three decision models, no graphics card: I ran 144M, 421M and 4.2B System One models on my own CPU and wrote down what each one said. The 144M model missed "I am cancelling on Friday"; the vague ticket produced three different answers. What a decision model is, how to run one with llama.cpp or Ollama, and how Kev, Nimble, Laya and Jev compare.
ai · 19 min · 2026-10-14




















