Zhongjin: Multi-modal reasoning helps to enhance the intelligence of driving abilities, and related main lines are worth attention.

The Golden Securities stated in the article that Google Gemini 2.5 was released in March, which can achieve multimodal fusion reasoning; in April-May, companies such as Horizon Robotics, SenseTime, and MiniMax successively released multimodal reasoning achievements. We believe that the significance of technological progress lies in: with the addition of multimodal thinking chains, the two main lines of multimodal and reasoning models are achieving architectural unity, and multimodal understanding capabilities are seeing a boost. Recently, companies like NIO and ideal have implemented multimodal reasoning in car-end interactions, and the integration and innovation of technology architecture are expected to continue bringing about the extension of application scenarios. The main line of multimodal reasoning is worth paying attention to.