Just tested, even Xin Qiji scored 666.

Author: Bai Jiao Heng Yu from Aofeisi, Quantum Bit

Image source: Generated by Wujie AI

Upon opening my eyes, WeChat confirmed:

It has officially integrated DeepSeek-R1, and it’s still the full version!

The usage path is very simple: open WeChat — click the search box — select "AI Search" in the middle of the page — choose deep thinking mode — and start using it~

The generated results also have social attributes, allowing sharing with friends or forwarding to Moments.

Previously, many people called for a one-click sharing feature for DeepSeek's answers, but their wish was unfulfilled. Before last night, almost all methods of sharing DeepSeek-generated results were screenshots (as it shows the reasoning process, the screenshots had to be particularly long).

Now, this demand has finally been realized by WeChat.

The public account "Consensus Crusher" calculated overnight, and currently, WeChat can support 50 million to 100 million simultaneous online users (simultaneous online, not DAU).

This number basically meets the usage of the first batch of DeepSeek users on WeChat and exceeds the simultaneous online users of ChatGPT.

Although it is still in grayscale testing, this is clearly Tencent's continued embrace of DeepSeek, and it is now truly hands-on and usable.

Good news, the Quantum Bit editor has been among the first batch to be in grayscale testing.

Great joy, shall we test it out? —

Testing WeChat's Integration with DeepSeek

Currently, WeChat's AI search offers two modes:

• Quick Answer: Generates brief answers, suitable for general simple questions.

• Deep Thinking: DeepSeek-R1 serves as the model support, providing long, comprehensive answers that are more detailed.

Therefore, to deeply experience R1, you need to select "Deep Thinking Mode".

First, let’s see if it supports basic online connectivity.

In the past few days, the Yushu robot has been performing again, updating its dance videos, which are more agile and stunning than those from the Spring Festival Gala, even astonishing foreigners.

I posed this question to WeChat: What dance is the Yushu robot performing in its latest dance video?

The answer received was:

Its response was smooth, and the output was quick. It first displayed its reasoning process, determining through online search that this is a progress from February 14, 2025, then considering the user's interest in the robot, it provided additional information, ultimately arriving at the following answer.

There is a small bug here, which is that there is no clear distinction between the reasoning process and the formal answer, and the formatting did not come out. However, this is likely due to the instability of the grayscale test, and it should not be an issue later.

From the 10 reference materials it attached, the first 7 came from public accounts. This also proves that this AI search is supported by WeChat's vast content ecosystem.

Alright, with basic online reasoning functionality in place, let’s increase the difficulty.

Previously, people used DeepSeek for various creative tasks, such as fortune-telling and character creation.

With that in mind:

Please write a verse in the style of Xin Qiji about WeChat integrating DeepSeek.

After the thinking process ends, it will hide the reasoning process and provide a formal answer, which can be expanded at any time by clicking.

The final answer given is as follows:

While creating the verse, it also provided annotations, which contained many hidden meanings.

The first half represents the significance of the event, with "Wisdom Opens the Starry Sky" having a dual meaning, while the second half reveals functional attributes, such as drawing on WeChat's ecological data.

After answering, it will recommend related questions or "ask another question".

In this way, clicking on a question or continuing to ask can lead to continuous answers (but it does not support continuous answers based on previous context).

Moreover, as long as you do not exit, all Q&A will be presented on the same interface.

Overall, due to the limitations of grayscale testing and the experience window (only supporting mobile), there are some areas that need improvement.

For example, the search is only supported on mobile, and the information shared with friends and Moments cannot be viewed on the computer.

In addition, the answers are not as diverse as the official ones, likely due to limitations in output tokens, and it cannot find historical Q&A or provide continuous answers based on context like the official version.

From this perspective, WeChat's integration with DeepSeek still presents a short, fast, and efficient search method, which aligns well with WeChat's attributes of immediacy and private social networking.

While the world is still struggling with AI entry…

What effect does WeChat's integration with DeepSeek have?

In a nutshell:

While global tech giants are still fiercely competing for AI entry, Tencent, with a seemingly light "ecological integration," has been dubbed by netizens as having "ended the AI search competition."

Does WeChat's move deserve this praise? Let’s sort it out together.

First, from a subjective experience, using DeepSeek-R1 on Tencent's Yuanbao is faster and smoother than using it on WeChat.

But that’s not important.

What matters is that WeChat's integration of DeepSeek-R1 provides a zero-learning-cost entry setup for the vast number of DeepSeek users that are hard to quantify.

In this way, 1.67 billion WeChat users (data sourced from the internet, as of October 2024) can directly access the strongest reasoning model in the Chinese domain without downloading a new app or registering a new account.

Secondly, the real killer feature lies in the fact that this move, while providing a simple and accessible entry, deeply integrates R1 with WeChat's ecological data resources and RAG (Retrieval-Augmented Generation) technology.

When users inquire about "how to book tickets for the Forbidden City," DeepSeek can not only search the entire internet but also, most notably, it can capture relevant information from public accounts, video accounts, and mini-programs, and can access the service interface of the official mini-program of the Forbidden City — this is truly Tencent's exclusive ecological resource.

Finally, the answers are clearly presented with added emojis, making it easy to read and highlighting key points, which adds a human touch.

As we have previously noted, Tencent's embrace of DeepSeek has become broader and more comprehensive.

Previously, Tencent Cloud, the knowledge application development platform Knowledge Engine under Tencent Cloud, the AI smart workstation ima, and the main AI application Yuanbao have all successively integrated DeepSeek-R1.

With such open-source success and Tencent's openness, what is the ultimate result?

It is that users can use the most convenient methods to widely incorporate useful AI into their mobile phones with the lowest learning costs.

The public account "Consensus Crusher" stated in its latest post that this move is a "turning point for 2C applications."

It also revealed some information:

△ Screenshot from the public account "Consensus Crusher" post "WeChat + Deepseek: A Turning Point for 2C Applications"

In the era of mobile internet, super applications are still leveraging their advantages today, and there are minds creating super applications.

In summary, starting today, the most commonly used application by Chinese users + the strongest domestic reasoning model + high-quality ecological data resources have been seamlessly integrated by WeChat and DeepSeek.

One More Thing

Yesterday, Perplexity launched Deep Research for free.

In terms of performance, it used the recently popular Humanity’s Last Exam to test AI reasoning capabilities, achieving an accuracy rate of 20.5%.

This outperforms Gemini Thinking, o3-mini, o1, DeepSeek-R1, and many other mainstream models.

In response, Perplexity's CEO publicly thanked DeepSeek:

This is because DeepSeek is open-source, cheap, and fast.

But Ultraman couldn't sit still—this morning, he called out on Twitter, directly targeting the Perplexity CEO.

He said:

ChatGPT (4o) is the best search product on the internet!

But now?

DeepSeek has not responded, and WeChat remains silent.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。