Ну от розробники моделі й не зацікавлені, щоб у користувача таке бажання виникало. Ну і не факт, що користувач почне розвиватися самостійно, а не піде жалітися ЧатуГПТ, який Клод козел.

A developer in Italy ran a 744 billion parameter AI model called GLM 5.2 on a laptop without a graphics card. The project, Calibri, streams only the active 10 GB of the model in RAM while loading the rest from the hard drive, enabling efficient offline operation without expensive GPU hardware. This breakthrough challenges NVIDIA’s GPU-dependent approach by showing that large-scale AI models can run on standard laptops. Calibri offers an accessible alternative for offline AI use, reducing costs and dependence on cloud services. By cloning the repository and running a simple setup, users can interact with the AI offline or deploy it as a local API endpoint, ensuring data privacy and zero token costs. The model download requires 370 GB of disk space, but afterward, the system runs smoothly without specialized hardware.