Requires 4gb of ram to run inference so iPhone14, iPadPro or Mac only. We don’t call them “small language models”.