On February 8, Dongfeng Motor announced that it has successfully integrated the full range of DeepSeek large language models. Independent brands such as M-Hero, eπ, Aeolus, and Nano will soon incorporate and deploy the technology in their vehicles.
Among them, the intelligent cockpit of the Dongfeng M-Hero 917 has already integrated the DeepSeek-R1 model, with an over-the-air (OTA) update expected in April 2025.
Through continuous customized model distillation and AI training, M-Hero owners will experience a significantly enhanced smart cockpit, featuring:
Faster voice recognition: Response speed improved from seconds to milliseconds.
More natural interactions: Improved semantic understanding and human-like responses.
Expanded off-road voice command library: More intuitive and scenario-specific commands for off-road driving.
DeepSeek's Growing Influence in the Auto Industry
The rise of DeepSeek is rapidly reshaping the tech landscape, particularly in the automotive sector.
On February 7, Voyah Auto announced the deep integration of DeepSeek into its vehicles. The Voyah Zhiyin will become the first mass-produced model in the auto industry to incorporate DeepSeek, with the new Voyah Dreamer also featuring the integration.
Previously, Voyah's intelligent cockpit had already completed the integration and deployment of DeepSeek's full model suite. The Voyah AI assistant will continuously integrate with the DeepSeek-R1 model in the cloud, enabling more precise, intelligent, and flexible interactions through enhanced voice control and scene-based recommendations.
Additionally, the Voyah Zhiyin vehicle system will continue refining the DeepSeek-R1 model through cloud-based integration and distillation. By leveraging DeepSeek, Voyah aims to not only optimize the in-car AI experience and accelerate technological iterations but also reduce the development and operational costs of smart cockpits.
Geely Also Joins the DeepSeek Race
On February 6, Geely Auto announced that its self-developed AI models have been deeply integrated with DeepSeek. According to official information, following the integration of DeepSeek-R1, Geely will apply distillation training to its Xingrui Vehicle Control FunctionCall model and on-device AI interactive models, leveraging both parties' strengths to accelerate AI adoption in smart vehicles.
Industry Insights: The Future of Smart Driving Competition
From these recent developments, it is clear that user experience, iteration speed, and cost efficiency will be the key factors in the 2025 intelligent driving race. At the core of these factors lie data, algorithms, and computing capabilities.
However, the direct use of DeepSeek for autonomous driving remains unlikely in the near future. The significant differences in data sources and application scenarios present major challenges. Additionally, stability issues in DeepSeek's current implementation further limit its feasibility for direct integration into autonomous driving systems.
