GAC Group to roll out DeepSeek-powered ADiGO SENSE AI model in H1 2025

Shanghai (Gasgoo)- On February 8, GAC Group released a real-world demonstration video showcasing its in-car AI assistant responding to voice queries powered by the DeepSeek-R1 model. This marks the deep integration of the group's self-developed ADiGO SENSE cloud-edge AI model with the DeepSeek-R1.

The enhanced ADiGO SENSE model, powered by the DeepSeek-R1, will be rolled out in the first half of this year, serving vehicles across GAC Group's HYPTEC, Trumpchi, and AION brands gradually.

GAC stated that it will continue to drive the innovative in the application of DeepSeek AI model in smart cockpits, scenario-based services, and user interaction experiences.

By leveraging DeepSeek-R1 through model distillation and fine-tuning, the ADiGO SENSE model strengthens both edge-side comprehension and cloud-based generative AI capabilities. This results in more accurate intent recognition, improved task distribution, and enhanced memory retrieval and association. Under the cloud-edge collaborative framework, the cloud-born AI model integrates DeepSeek-R1, enabling more advanced reasoning and reflection capabilities to deliver highly personalized services.

This integration transforms the car from a "command-driven tool" into a "companion that proactively predicates demands," redefining smart cockpit interactions. The DeepSeek-R1 algorithm breakthroughs will significantly enhance GAC Group's intelligent system's cognitive and reasoning abilities, bringing four key advancements: 

1. Ultra-low latency interactions– The lightweight design of DeepSeek-R1 enables complex AI models to run efficiently on in-vehicle systems, reducing voice assistant response times to the millisecond level by leveraging both local user data and cloud-based resources.

2. More natural and seamless interactions– The deeply integrated cloud-edge model supports multi-modal inputs, including voice, text, images, and driver behavior, enabling more precise understanding of user intent and environmental context for smoother interactions.

3. Recognition of implicit user needs– By combining vehicle status, historical user behavior, and real-time environmental data, the system builds a dynamic scenario knowledge graph for deep contextual reasoning, identifying not only explicit but also implicit user demands.

4. Intelligent execution of scenario-based services – The ADiGO SENSE AI model accurately interprets ambiguous user commands and connects with over 2,000 in-vehicle service interfaces, enabling seamless system integration and proactive assistance.