Xiaomi has officially released the 7B parameter multimodal large model "Xiaomi-MiMo-VL-Miloco-7B-GGUF" on Hugging Face and GitHub today, and launched the intelligent assistant "Xiaomi Miloco" based on this model.
The system can recognize user activities (such as gaming, fitness, reading) and gestures (victory gesture, thumbs up, etc.) in real-time through Mi Home cameras, automatically connect to smart home devices such as lights, air conditioners, and music, and is compatible with the Home Assistant protocol. Miloco uses a non-commercial open-source license, and users can deploy it with one click on Windows or Linux hosts equipped with NVIDIA GPU and Docker environment.
Official examples show that the system automatically turns on the desk lamp in reading scenarios, adjusts the air conditioner according to whether the user is covered in sleep mode, and generates voice comments based on the user's clothing style when entering home. These are default workflows. Xiaomi stated that model weights and inference code are publicly available, but intellectual property rights are reserved, and commercial use is prohibited.
