Baidu reveals significant investments in AI applications and cost-effective models at the Wuhan developer conference

Baidu invests in affordable AI models at Wuhan conference.

: Baidu announced major AI investments at the Wuhan Create 2025 conference, highlighting affordable AI models like Ernie 4.5 Turbo and X1 Turbo. CEO Robin Li emphasized the potential of multimodal models and the importance of strategic application development to stay relevant. Competitors like DeepSeek face pricing challenges as Baidu undercuts them significantly. Tech partners, including NVIDIA and Intel, attended the event, which featured prize money to encourage AI innovation.

At the recent Wuhan Create 2025 conference, Baidu made significant announcements about their investments in AI, focusing on cost-effective models and smarter applications. CEO Robin Li addressed a crowded room of developers, highlighting the dual nature of foundational models, which can both enable new app developments and risk app obsolescence if not strategically applied. The key, according to Li, is to identify pertinent scenarios, select appropriate models, and understand fine-tuning.

Baidu launched two new large language models, Ernie 4.5 Turbo and X1 Turbo, strategically targeting their AI rival DeepSeek. These models offer superior multimodal reasoning capabilities at a fraction of the cost, with Ernie 4.5 Turbo priced at 40% less than DeepSeek’s equivalent and X1 Turbo costing just 25% of its competitor’s rates. Ernie 4.5 Turbo is priced at RMB 0.8 ($0.11) per million tokens for input; X1 Turbo has an input rate of RMB 1 ($0.14).

Li strongly advocates for a future dominated by multimodal models, as opposed to purely text-based ones, suggesting that handling diverse data types like text and images will become the industry standard. The launch of Xinxiang, a new AI agent app by Baidu, shows their ambition in the autonomous agent space, competing with platforms like Manus by employing the Agent Use protocol to execute tasks directly.

Baidu is also focusing on expanding its AI ecosystem by inviting third-party developers to build applications on its search engine and adopting the MCP (Model Capability Protocol) standard. This move intends to make its platform more developer-friendly, aiming to outperform competitors like Alibaba Cloud and Tencent Cloud who have already integrated MCP into their ecosystems. Li emphasized that while many models will emerge, app supremacy will ultimately determine competition outcomes in the AI sector.

The conference attracted significant interest from tech giants such as Intel and NVIDIA and included emerging players like Unitree Robotics and Leapmotor. A notable announcement was the launch of a new round of Baidu’s AI app competition, offering RMB 70 million ($9.6 million) in prize money to drive AI innovation and development within China's burgeoning tech industry.

Sources: Baidu, Unitree Robotics, Technode

innovationInvestment