March 20, 2010 - Hugging Face has launched its latest iOS app, HuggingSnap, which allows users to ask AI to generate visual descriptions directly on the device side without relying on cloud servers. The app is based on the lightweight multimodal model smolVLM2 (with parameter sizes ranging from 256 million to 2.2 billion), which performs all calculations locally and avoids the need to upload data to the cloud to ensure privacy. Optimized for mobile devices, smolVLM2 efficiently handles graphical tasks (e.g., image/video analysis) but is slightly less accurate than the cloud...