Put two parts side by side — specs, price, and for graphics cards, which local AI models each one can run.