At the 2026 Yunqi Conference, XPENG and Banma announced a collaborative automotive AI cockpit solution. This innovation leverages an on-device large model capable of local deployment, facilitating advanced computing capabilities in vehicles.
Key Highlights
- Joint Solution: Combines XPENG’s Turing AI chip with Banma’s AutoOmni multimodal large model, supporting multi-turn conversations and Q&A.
- Performance: The Turing AI chip offers 750 TOPS (Tera Operations Per Second) computing power, tailored for automotive and robotics applications.
- Architecture: The “on-device first, cloud-assisted” model allows core AI functions to operate locally, with the cloud handling less frequent tasks like information updates.
Technology & Features
- Computing Foundation: Utilizes XPENG’s Turing AI chip and supports 30B-class models for functions like voice interaction and complex task planning.
- Real-Time Processing: Ensures real-time inference within the vehicle, enabling seamless interactions without relying heavily on cloud services.
- Data Privacy: Processes cockpit data locally to address connectivity issues and minimize data transfer needs, particularly beneficial in international markets.
Strategic Goals
The partnership aims to equip automakers with a robust “chip + AI OS + large model” ecosystem, enhancing in-vehicle experiences. Banma has already partnered with over 60% of automakers for its cockpit AI solutions, indicating strong industry support.
Conclusion
This collaboration signifies a pivotal step towards integrating high-performance AI into vehicles, enhancing user interaction, safety, and overall smart driving experiences.
