SenseTime "SenseAuto Kaiwu" World Model, Ranked No.1 in China!

On September 24, 2025, Frost & Sullivan, a leading research institute, released the “White Paper on the Development of China's World Models in 2025”.

In the report, Frost & Sullivan pointed out that there are over 10 players in China's world model sector, among which SenseTime's "SenseAuto Kaiwu" World Model ranks first in the comprehensive evaluation across three aspects: technological products, application capabilities, and ecosystem development.

As a type of generative AI model, world models can understand the dynamic laws of the real world by building internal representations, while generating video content using input information such as text, images, videos, and motion data. By learning representations from sensory data and predicting dynamic characteristics like motion, force, and spatial relationships, world models can understand the physical properties of real environments, and simulate, guide, and implement decisions by generating environments and actions.

World models have demonstrated great application potential in cutting-edge fields such as autonomous driving and robotics. The report notes that over 80% of autonomous driving algorithms use world models for auxiliary training. By automatically generating annotated image and video data, as well as multi-modal, cross-temporal consistent scenarios that require no extensive manual design, world models can help enterprises save nearly 50% of costs and improve efficiency by approximately 70%.

Regarding the currently booming embodied intelligence sector, the report believes that in the future, world models will become the "cognitive core" of embodied intelligence. World models can provide a high-quality, low-cost, and scalable synthetic data generation path for embodied intelligence, solving the current data bottleneck. In addition, building a one-stop development platform for embodied intelligence with world models at its core can provide integrated toolchains and environments, reshaping the development paradigm of embodied intelligence.

From a global market perspective, well-known manufacturers such as SenseTime SenseAuto, Huawei, NVIDIA, Google, Meta, and Wayve have all built unique world model capabilities and related products based on their own strategies and technical advantages in different dimensions. The report points out that product technology, application capabilities, and ecosystem development have become key indicators of the comprehensive strength of various world models.

As the industry's first mass-produced and interactive world model, the "SenseAuto Kaiwu" World Model has empowered the fields of intelligent assisted driving and embodied intelligence. SenseAuto has built the first generative world model product platform in the assisted driving field and "WorldSim-Drive", the industry's largest-scale generative driving dataset.

With the "SenseAuto Kaiwu" World Model as its core engine, SenseAuto has built the Wuneng Embodied Intelligence Platform, which enables visual perception, precise navigation, and multi-modal interaction. Supported by end-side and cloud-side computing power, the platform allows intelligent agents to independently understand and act in real environments.

Looking to the future, the "SenseAuto Kaiwu" World Model is transforming the creativity of AI into productivity, building a bridge for AI to interact with the real world.