General - Runtime: Use CUDA if available (NVIDIA GPU); if not, use Vulkan and offload as many model layers to the GPU as possible to significantly improve text generation speed.
Steam Edition - Guard: You can disable the option to keep the Guard model loaded at any time (to reduce loading time), in favor to further reduce the ram usage of the App while no other Textmodel is loaded.
General: - You can minimize the app to the System Tray. This also allows you to switch between Widget Spaces via the System Tray Icon.