At the 2025 World Artificial Intelligence Conference (WAIC), Baidu showcased multiple innovative achievements and the latest developments in the field of artificial intelligence.
During the conference, Baidu announced that its Apollo Go, PaddlePaddle Deep Learning Platform, and Baidu Intelligent Computing Cluster have been selected for the China AI Industry Innovation Exhibition. Among them, Apollo Go not only served as an exhibition item but also acted as a shuttle vehicle for the conference, demonstrating the mature application of its autonomous driving technology. Currently, Apollo Go has provided over 11 million travel services worldwide, with a safe driving mileage exceeding 170 million kilometers, and is accelerating large-scale deployment in places such as Dubai and Abu Dhabi. It has also partnered with Uber to integrate autonomous vehicles into the global mobility network.
The PaddlePaddle Deep Learning Platform, as China's first independently developed, feature-rich industrial-grade deep learning platform, integrates core components for deep learning training and inference frameworks, serving 670,000 enterprises, creating 1.1 million models, and promoting the intelligent upgrading of thousands of industries. As of April 2025, the number of WENXIN developers on the PaddlePaddle platform has reached 21.85 million, reflecting its broad influence within the developer community.
In addition, Baidu successfully lit up China's first fully self-developed 30,000-card intelligent computing cluster. This cluster can simultaneously support the full-scale training of multiple trillion-parameter large models and support hundreds of customers to perform fine-tuning of hundred-billion-parameter large models, effectively enhancing computing power supply and reducing costs.
In the field of digital human technology, Baidu Hui Boshi released the next-generation digital human technology NOVA, which achieved three core breakthroughs: based on the WENXIN Large Model 4.5 to generate high-quality scripts and drive multimodal collaboration; equipped with an AI brain that can receive live broadcast data in real time and make autonomous decisions; and an efficient replication capability that can complete voice, emotion, and action replication with only 10 minutes of real person samples. The NOVA technology helped Luo Yonghao's digital human live stream room achieve a high GMV, and it is expected to be open to the entire industry in October, marking the entry of super head主播 ability replication into the era of large-scale mass production.