WeRide Launches WITT AI Model to Boost Autonomous Driving Data Efficiency — WeRide Inc.

Event summary

WeRide introduced WITT (World Intelligence Toward Truth), a Physical AI Cognitive Foundation Model designed to extract and verify atomic physical facts from real-world driving data.
WITT reduces token costs by up to 98% and delivers up to 200x greater data-processing efficiency compared to general-purpose AI models.
The model is integrated into WeRide's Physical AI flywheel, which includes the company’s proprietary simulation model GENESIS.
WeRide operates a fleet of over 3,000 autonomous vehicles across more than 40 cities in 12 countries.

The big picture

WeRide's launch of WITT addresses a critical industry challenge: transforming vast amounts of real-world data into trusted learning signals for AI training. As autonomous driving moves from research to large-scale commercial validation, the efficiency and accuracy of data processing become paramount. WeRide's Physical AI flywheel, combining WITT with GENESIS, positions the company at the forefront of this technological shift.

What we're watching

Data Efficiency: How WITT's ability to reduce token costs and improve data-processing efficiency will impact WeRide's operational margins.
Commercial Deployment: Whether WeRide can sustain its large-scale commercial deployment of L4 autonomous driving systems across multiple markets.
AI Integration: The pace at which WITT's integration with GENESIS will enhance the overall performance of WeRide's autonomous driving solutions.