Sina Technology tested multiple questions on WeChat and found that most of the reference materials attached to the answers provided by DeepSeek came from WeChat public accounts. Interestingly, when asked on WeChat "Who is the biggest loser when WeChat connects to DeepSeek?", the answer given by DeepSeek is: Baidu.


Text | Sina Technology Liu Lili

Baidu becomes the biggest loser?

The market generally believes that WeChat's access to DeepSeek marks the combination of super traffic portals and top open source models, which will improve the level of search intelligence and accelerate industry reshuffle in the short term, and may reshape the "search as a service" ecological pattern in the long term.

In this carnival, DeepSeek bluntly said: Baidu is the biggest loser.


(Ask on WeChat: Who is the biggest loser when WeChat accesses DeepSeek?)
(Ask in Wen Xiaoyan: Who is the biggest loser when WeChat connects to DeepSeek?)

Baidu's search market share continues to decline, having dropped from 86.82% in 2021 to 60% in 2024, while platforms such as Xiaohongshu and WeChat have gradually seized user habits through AI search.

DeepSeek believes that after WeChat integrates DeepSeek, users can obtain accurate answers directly within the ecosystem, further weakening Baidu's reliance on traditional search. Moreover, WeChat has a large amount of content such as public accounts and video accounts. Combined with DeepSeek's semantic understanding and personalized recommendation capabilities, it can activate long-tail content and implement "search as a service" (such as directly calling mini programs). Baidu lacks a similar content ecological closed loop and is difficult to compete with.

"What does this wave of WeChat mean? It means standing aside." Some analysts bluntly said, "If Liang Wenfeng flipped the closed-source model table during the Spring Festival, the path chosen by WeChat, which has more than one billion users around the world, is to cover the board, press the nails, and wait to eat."

The above-mentioned people believe that although Baidu's Wenxin model has been announced to be completely free from 0:00 on April 1, it may not help the problem. Platforms such as WeChat have achieved technological leaps through low-cost access to DeepSeek, while Baidu is bound to gradually fall behind in data integration and user experience.

DeepSeek’s view on this issue is, “Baidu is facing multiple impacts on technology, ecology, and user habits. If it cannot transform quickly, it may gradually lose its competitiveness in the new era of AI-driven search.”

Is the WeChat version of Deepseek better than Deepseek itself?

In fact, so far, the content of WeChat official accounts and Xiaohongshu has basically formed a closed loop. Some excellent content in the WeChat ecosystem is not retrieved by other external search engines and large models. Moreover, the content of WeChat official accounts is several times the volume of the content of Xiaohongshu, so the quality is naturally much higher than that of public searches.

It is reported that after WeChat is connected to the DeepSeek-R1 model, the answer content will give priority to integrating WeChat public accounts, video accounts and other information sources within the WeChat ecosystem to strengthen the closed loop of content within the platform.


In multiple tests, it was found that most of the reference materials attached to the answers provided by DeepSeek come from WeChat public accounts.


After WeChat is connected to DeepSeek, what benefits will it bring to WeChat content creators?

DeepSeek's response to related questions is that after WeChat is connected to DeepSeek, the core competitiveness of creators will shift to "data-driven rapid iteration capabilities" instead of simply pursuing hot hits. Its long-term value includes: the creation cycle is compressed from 7 days to 12 hours, and the conversion efficiency of the entire process is increased by 200%+; the AI ​​recommendation mechanism allows historical content to be rediscovered, and the content of public account articles and video accounts to gain secondary exposure; AI assists in generating copywriting and scripts, reducing creation costs, and novices can quickly produce high-quality content.

For creators, the most attractive thing is the "demand insight" through AI, which predicts user needs and precise content. DeepSeek claims that it can analyze comments and private domain chat records through NLP to identify users' hidden needs, and the efficiency is three times faster than traditional data analysis; combined with WeChat index and DeepSeek trend prediction, it can identify explosive topics 48 hours in advance.

This is undoubtedly a pain point for traffic-hungry creators.

In addition, WeChat is connected to the DeepSeek-R1 model and supports direct sharing to Moments or group chats, which greatly enhances social attributes. The high-quality content on the WeChat platform is equivalent to giving R1 a massive knowledge base, and the quality of answers will also be improved.

Therefore, some people believe that the WeChat version of Deepseek may be much easier to use than Deepseek itself.