Baidu launches the "O Plan": from search to "intelligent agent scheduling"

Author | Zhou Zhiyu

On the eve of the Spring Festival in 2026, the atmosphere in China's internet sector is more intense than in previous years.

On February 10, Wall Street News learned from insiders at Baidu that the company has recently quietly launched a project codenamed "Plan O." Unlike previous departmental collaborations, this time the search and cloud teams are working together, indicating that the project breaks down departmental barriers, with MEG and ACG collaborating.

Other informed sources revealed that this collaborative project is related to the Baidu App. Baidu aims to focus on users' daily scenarios, using the Baidu App as a hub, leveraging Wenxin Assistant to mobilize Baidu's internal ecosystem and partner services to address users' actual needs.

This series of actions sends a strong signal. Baidu realizes that the competition for AI entry points is no longer just a front-end UI competition, but a comprehensive battle involving computing power costs, inference speed, and data scheduling capabilities.

In simple terms, the UI is just a dialogue box that anyone can create; but what truly creates generational gaps is the underlying scheduling logic and computing power ledger. When a user presents a complex demand, the system needs to instantly mobilize services from the entire ecosystem. If the scheduling capability is poor, AI will just be a chatty facade that cannot truly solve problems; and if computing efficiency is low, the cost of a single response will be exorbitant, making it commercially unviable.

The launch of "Plan O" by Baidu is not without precedent. Baidu founder Robin Li clearly stated in an internal sharing session at the beginning of 2026 that the purpose of training is inference.

This means that inference efficiency will determine the generational competition in search. If Baidu cannot connect the underlying layers of search and cloud to significantly reduce the inference cost of single intent analysis, then the so-called "AI search" will turn into a losing business.

In the past, the search team guarded trillions of query traffic, which was a cash cow and a stable resource; while the large model team represented the future, which is incremental. Over the past two years, the relationship between Wenxin Assistant and Baidu Search has been two parallel logics. However, in 2026, with the launch of "Plan O," this state has come to an end.

Baidu Search is no longer a display hall but the knowledge base of Wenxin Assistant. Wenxin Assistant has become the driver holding the steering wheel.

Wenxin Assistant is the front end, responsible for perceiving user emotions and vague intentions; search is the posterior, responsible for extracting authenticity verification from massive real-time data.

This model means that Baidu is no longer entangled in the short-term fluctuations of search advertising click-through rates (CTR). Because if Baidu does not undergo self-revolution, ByteDance's Doubao and Alibaba's Qianwen will impact the traditional search advertising model with an interaction experience of "low advertising interference, directly providing results."

The Spring Festival of 2026 marks the "D-Day" for China's large model industry.

As Baidu, Alibaba, Tencent, and ByteDance collectively enter the red envelope battle, this war has long surpassed simple marketing strategies, evolving into a life-and-death battle over the "default entry point."

In the competition for AI entry points, major companies have taken completely different paths based on their respective genes.

Doubao, as a partner of the Spring Festival Gala, has a logic of extremely powerful content distribution. ByteDance does not teach users how to think; instead, it uses short videos, social chains, and extreme UI interactions to make the AI assistant a new toy for users to pass the time. Doubao's high DAU is its core weapon, attempting to integrate AI into daily life through high frequency and low frequency Qianwen has thrown out 3 billion for "Spring Festival hospitality," and its intention is clear—AI must be able to generate transactions. Alibaba deeply binds AI with local life and e-commerce, allowing users to order milk tea or return items with just a phrase. Alibaba's AI entry is essentially the universal remote control of its vast business empire.

Yuanbao relies on WeChat, leveraging familiar relationship chains. Yuanbao is not in a hurry to become independent; it is more like an intelligent upgrade of the WeChat ecosystem, internalizing AI as a tool for social interaction through the distribution of red envelopes.

Now, Baidu has also found its positioning. It aims to solve users' clearly defined, even complex problems, relying perhaps on the deep content accumulated through search and its full-stack AI capabilities.

As Wenxin Assistant handles intent planning, search is responsible for knowledge verification, and Baidu Intelligent Cloud manages low-cost, high-concurrency rapid inference execution, a prototype of an AI intelligence dispatch center may emerge.

By 2026, the so-called "hundred model war" will have completely dissipated. The industry will enter an extremely brutal period of entry fragmentation.

Baidu, relying on search scenarios, has a natural advantage in the AI entry competition, but it will inevitably face the challenge of independent app diversion.

Search itself is the most natural soil for Q&A. If the "O Plan" can successfully transform search into an intelligent agent distribution platform, Baidu may also transform into an operator in the AI era.

In the past, discussions about entry focused on traffic. Now, discussions about entry focus on mental occupation. Whoever can make users speak directly to their phones at the first thought of a need, rather than opening an app to search, will hold the future pricing power.

Baidu's "O Plan" is a gamble, betting that users will develop a stickiness to in-depth answers. Meanwhile, ByteDance and Alibaba are betting on satisfaction and convenience. This differentiated showdown will be determined in every red envelope circulation and every voice interaction during the Spring Festival.

This is not just a competition about technology; it is also a game of courage. There is no turning back in front of the AI entry