2 1 Shi Ji Jing Ji Bao Dao·2025-04-27 01:15Core Insights - Baidu is focusing on developing large models and applications to create a thriving ecosystem for AI development [2][3] - The newly introduced models, Wenxin 4.5 Turbo and Deep Thinking Model X1 Turbo, feature multi-modal capabilities, strong reasoning, and low costs [2][4] - The cost of using Wenxin 4.5 Turbo is significantly reduced, with input prices at 0.8 RMB per million tokens and output prices at 3.2 RMB, representing an 80% decrease compared to its predecessor [2][6] - The X1 Turbo model further reduces costs, with input prices at 1 RMB and output prices at 4 RMB, achieving a 50% price drop compared to the previous model [2][6] Application Development - Li Yanhong emphasized the importance of applications, stating that without them, models and chips hold no value [8] - Multiple AI applications were launched, including a highly persuasive digital human and a general super-intelligent agent app, covering popular sectors like digital humans and code intelligence [5][6] - The highly persuasive digital human can generate scripts and adjust expressions and actions in real-time, enhancing the live-streaming experience [6][7] Cost Efficiency - The cost of creating digital humans has decreased to one-third of last year's levels, with monthly costs for businesses now under 1,000 RMB [6][7] - The technology behind digital humans is rapidly evolving, indicating potential for further cost reductions in the future [7] - Baidu aims to empower small businesses and merchants by providing them with digital human capabilities that rival real humans, thus enhancing productivity [7] Developer Support - Baidu launched the "AI Open Plan" to support developers by providing diverse content and service distribution mechanisms [7][8] - The MCP server discovery platform was introduced to index high-quality servers available in the market, facilitating easier access for developers [7][8] - Li Yanhong encourages developers to innovate and create applications, highlighting that applications will dominate the future of AI [8]