Recently, Baidu founder Robin Li elaborated on the reasons why the company decided to open source Wenxin Mockup 4.5 series at Baidu's fourth quarter and annual financial report conference call. This decision not only demonstrates Baidu's leading position in the field of artificial intelligence, but also reflects the company's profound insight into future technological development.
Li Yanhong pointed out that Baidu has learned valuable experience from the DeepSeek project that the best open source model can greatly promote users' adoption and recognition of the model. He explained that when a model is open source, people's curiosity drives them to try, and this natural behavior of exploration helps the model to be more widely used in real scenarios. Through open source, Baidu hopes to accelerate the market penetration of Wenxin Mockup 4.5 series and promote its application in more industries.
Li Yanhong emphasized that the Wenxin Mockup 4.5 series is the most advanced model ever. The company hopes to make it easier for users and customers to try and experience its excellent performance through open source. He further stated that Baidu decided to open source model because of its firm confidence in its technological leadership. This confidence comes from Baidu's continuous R&D investment, technological innovation and its unique advantages in UM4AI stack capabilities over the years, making Baidu one of the few companies in the world with this capability.
At the same time, Robin Li also clearly pointed out that whether it is open source or closed source, the real value of the basic model lies in whether it can effectively solve practical problems and achieve large-scale application. He emphasized that Baidu's open source Wenxin Mockup 4.5 series aims to promote the application of this model in a wider range of fields so as to fully realize its potential value. Through this measure, Baidu can not only enhance its technological influence, but also make important contributions to the development of global artificial intelligence technology.