Li Auto Unveils MindVLA-o1 Model

[Technical Breakthrough] At NVIDIA's GTC conference, Li Auto announced its next-generation autonomous driving foundation model, MindVLA-o1.

Core Development: MindVLA-o1 Addresses the Bottleneck in Deploying Physical AI

Zhan Kun, head of Li Auto’s foundational models, shared the stage with representatives from NVIDIA and Tesla, highlighting that the current bottleneck in autonomous driving lies in the lack of understanding of the real 3D world. Traditional modular approaches are rigid, while pure end-to-end models become "black boxes." MindVLA-o1 aims to combine the strengths of both paradigms.

Key Metrics: RMB 102.1 Billion in Cash Fuels Long-Term Investment

By the end of 2025, Li Auto’s cash reserves reached RMB 102.1 billion, providing ample resources for physical AI R&D. The company emphasized that many core technical leadership roles are now held by post-90s and post-95s talent.

Strategic Foundation: Backcasting Technology Roadmap from the Endgame

Li Xiang stated that the ultimate goal of the automotive game is "silicon-based life." Li Auto is leveraging physical AI as a fulcrum to reconstruct the foundational logic of autonomous driving, driving the industry’s transition from rule-based systems to cognition-driven systems.