Hugging Face launched the D1 family of open multimodal decision models optimized for edge devices. These models process visual and textual inputs to trigger specific actions without relying on cloud infrastructure. This reduces latency for real-time robotics and IoT applications.