The first quarter of 2025 has witnessed an explosive surge in the AI application arena. The question on everyone’s mind is: Which AI app has captured the global spotlight in Q1? The answer is critical, as it will shape the future of AI applications and redefine the competitive landscape for major corporations, potentially unveiling the next AI unicorn.
According to data from QuestMobile, the number of active users of AI-native apps reached 240 million by February 2025, nearly doubling from January. This surge was primarily triggered by the widespread acclaim surrounding DeepSeek-R1, an inference model launched earlier this year by AI firm DeepSeek.
The market’s enthusiasm was further fueled by the introduction of Manus, touted as the world’s first general-purpose AI Agent product, in early March. Despite still being in the beta phase, Manus’s invite codes were being resold on secondary markets for thousands of dollars due to overwhelming demand.
This frenzy has not only ignited the capital markets but has also fostered a collective sentiment that we have entered the ‘Year One’ of AI applications. The key question is, which AI application has emerged as the most popular globally in the first quarter? The answer to this question reveals the future trajectory of AI applications, impacts the strategies of leading companies, and may even hold the key to the next AI unicorn.
To address this, we’ve compiled data from prominent AI leaderboards such as AI Product榜 (AI Product Ranking), Xsignal, AIGCRank, and 新榜 (New Rank), integrating key metrics like monthly active users (MAU), daily active users (DAU), and download numbers, along with insights from industry professionals. This analysis resulted in a list of the top 20 global AI applications and top 10 domestic AI applications for the period of January to March. Our key findings are as follows:
- Existing AI application rankings are fragmented, with inconsistencies in data sources, ranking criteria, and results. A comprehensive view requires aggregating multiple lists.
- The top AI applications are relatively stable, with ChatGPT, 夸克 (Quark), 豆包 (Doubao), and DeepSeek consistently holding the top four positions.
- The influence of Chinese AI applications is growing, with Chinese and foreign applications achieving roughly equal representation in the global top 20.
- Market reception varies significantly by AI application type: chatbot, AI companion, AI photo editor, AI office assistant, and AI video generator.
- AI application rankings are not solely determined by technology but also by marketing capabilities and feature integration.
Decoding the Diverse AI Rankings
Before ranking AI applications, it’s crucial to define the criteria and their relative importance. However, reviewing multiple rankings reveals a confusing trend: the same application can be ranked vastly differently across various lists.
This inconsistency stems from two main issues:
First, ranking methodologies differ significantly.
Some lists aggregate data from both the APP and Web versions of an application to generate rankings, while others separate them into APP and Web rankings. The metrics used also vary, primarily including monthly active users (AI Product Ranking, Xsignal), daily active users (AIGCRank), and downloads (New Rank). Some lists use a combination of these, while others rely on a single metric.
Second, data discrepancies exist even for the same application and metric across different rankings.
For instance, AI Product Ranking and Xsignal both primarily use monthly active users to rank AI applications, but their data for each APP differs. According to AI Product Ranking, DeepSeek’s APP had 61.81 million MAU in February, while Xsignal reported 145.51 million.
According to a senior data practitioner, the discrepancies are due to differences in data sources. The AI rankings generally rely on two types of data sources. One type is self-monitoring, where companies are data-monitoring companies that develop AI application rankings after AI becomes popular. The other is to purchase data from third-party platforms and summarize it into lists.
The third-party purchase leads to many data sources not being first-hand, and the reliability of the data cannot be evaluated. The practitioner added that the AI industry is new, and different data monitoring companies have not yet developed a mature calculation system for AI applications, which can lead to data discrepancies.
Another ranking practitioner admits they purchase AI application data from third-party data companies. They have found significant differences in the data during the organization process, which has led them to hesitate in choosing data dimensions. After comprehensive consideration, they finally changed from the commonly used monthly active user dimension to daily active users to rank.
Despite these inconsistencies, a consensus has emerged within the industry, supported by multiple data practitioners: the importance of data dimensions ranks as monthly active users > daily active users > downloads > other.
‘If an application consistently ranks high across different lists, it’s safe to assume it’s performing well,’ one practitioner stated. Regarding the specific data discrepancies, practitioners recommend focusing on overall trends rather than delving too deeply into the details.
Based on these insights, we primarily used monthly active user data from various lists for our rankings, averaging the data when discrepancies occurred. Many applications have both APP and Web versions, but the APP version often indicates higher user engagement and active usage, so we mainly focused on APP data. We have also included data on APP variations to provide a relatively comprehensive frame of reference.
Global Competition: Five Popular App Categories, Education Tools Emerge as Dark Horses
First, let’s examine the top 20 AI applications worldwide in a single month.
In January, ChatGPT maintained a dominant lead in the global AI application rankings. Its advantages include its early start and advanced technology. It was the first AI application that became widely popular and now has a significant lead in MAU and monthly downloads, with 349.41 million and 695 million, respectively, far exceeding 夸克’s 143.4 million and 5.94 million.
Following closely are AI applications from major domestic companies, with ByteDance’s 豆包 and the new dark horse DeepSeek ranking third and fourth, respectively. Unlike other AI applications that have been regular fixtures on the lists, DeepSeek became a hit immediately. It had been unknown until before the Spring Festival, when it rose directly to the top of the list with its DeepSeek-R1 model.
Among the ‘AI Six Little Dragons,’ 月之暗面’s Kimi ranked eighth, and 智谱’s 智谱清言 ranked eighteenth. Baidu’s 文小言, 360’s 纳米搜索, and 科大讯飞’s 讯飞星火 all ranked outside the top ten.
Two major events occurred in the AI application market in February: the domestic AI applications began to access DeepSeek, and the international AI applications updated their products to respond to the impact of DeepSeek. These changes affected the rankings for that month.
The biggest changes were that Tencent’s AI chatbot 元宝 and American AI star company Anthropic’s Claude jumped from outside the top twenty to tenth and sixteenth, respectively. 元宝 benefited from the announcement that it would access DeepSeek on February 13. AI Product Ranking showed that its MAU for that month was 13.12 million. Claude launched a significant update in February, adding its ‘most intelligent model to date,’ Claude 3.7 Sonnet, and introducing hybrid reasoning capabilities for the first time, earning the market’s expectations.
In March, although an AI Agent application called Manus, which was described as ‘comparable to DeepSeek and shocking to Silicon Valley,’ appeared, it was also developed by a domestic startup team, but because the product only supported invitation codes, most users could not experience it, and it did not make the top twenty. Therefore, the AI application rankings in March did not change much, except that 元宝 further advanced from tenth to fifth. 文小言 also accessed DeepSeek in March, and its ranking rose from seventeenth in February to thirteenth.
It is worth noting that some APP in vertical fields have also appeared at the top of various lists. For example, the Xsignal list shows that the AI writing tool AI创作狮 had 3.72 million MAU in February, an increase of 20011.43% month-on-month. The AIGCRank list shows that 作业帮’s 快对AI and 猿辅导’s 小猿AI, these AI education tools,凭借高日活,都首次上榜国内应用榜单前二十.
In general, ChatGPT, 夸克, 豆包, and DeepSeek have ‘dominated’ almost all of the lists. In addition, we can draw some conclusions:
First, in terms of application type, the popularity of AI applications from January to March is: chatbot > AI companion > AI photo editor > AI office assistant > AI video.
The top four on the list are all in the chatbot field, followed by AI companions (AI emotional support) represented by Talkie AI and 星野, and AI photo editing tools like Remini. AI office tools like DeepL and Notion AI are ranked relatively lower.
A practitioner stated that chatbot are positioned as general-purpose AI tools, which naturally have high popularity, and major manufacturers are also focusing on this direction. AI companion meet the current demand of young people for emotional support and social interaction. As for photo editing and office AI applications, they are closely related to people’s daily life and work.
In contrast, although manufacturers such as ByteDance, Kuaishou, and MiniMax have invested heavily in the AI video track, C-side acceptance is not as good as other types, mainly because there is a certain barrier to entry. As for vertical App such as AI consultation and AI programming, they are still in the market cultivation period, and acceptance needs to be improved.
In the five popular AI fields, at least one representative product has emerged.
For example, in the chatbot field, ChatGPT is the strongest and is growing steadily, with downloads in March far exceeding those in January. According to a practitioner, ChatGPT’s MAU and downloads will continue to grow based on its current level of technological iteration. The competition in this track in China is even more intense, and 夸克, 豆包, and DeepSeek are not significantly separated. In the AI companion track, Talkie AI has the highest monthly activity, which is developed by domestic company MiniMax specifically for overseas markets.
Second, the competitive landscape between China and foreign countries.
More than one practitioner believes that the development speed of domestic AI applications is very fast, and the gap between China and foreign countries is narrowing. Among the top twenty applications in the entire Q1, the number of domestic and foreign applications is basically the same, and Chinese App occupy four of the top five, and the fastest growing is domestic 腾讯元宝.
This is partly due to more and more people actively embracing AI, and partly due to major manufacturers continuing to increase investment in AI application research and development and marketing efforts, as well as breakthroughs by some entrepreneurial teams such as Deepseek. The suspense of this global AI race is still escalating.
Domestic Battle: Marketing War Begins, 元宝 and 即梦 are the Biggest Variables
Let’s take a look at the situation in China.
The top ten AI applications in January were basically occupied by three major camps, including ‘Big Factory Series’ Alibaba 夸克, ByteDance 豆包, Baidu 文小言, 360 Nano AI Search, 科大讯飞 讯飞星火; ‘AI Six Little Dragons’ 月之暗面 Kimi, 智谱 智谱清言; and ‘New Dark Horse’ DeepSeek.
Their main battlefield is concentrated in the chatbot field, with the Big Factory Series performing the most prominently, with 夸克 and 豆包 occupying the top two. DeepSeek is ranked third.
Two other players in the top ten do not belong to the chatbot track, ByteDance’s 猫箱 and Mini Max’s 星野. They are both AI companion products and have very similar data in all dimensions, and no clear winner has emerged yet.
The pattern changed significantly in February, with ByteDance’s AI video generation tool 即梦 and Tencent’s 元宝 making significant advances, ranking ninth and fifth, respectively, while 讯飞星火 and 智谱清言 disappeared from the top ten.
It is not difficult to see that the domestic chatbot track still lacks uniqueness, and whether it can be listed is strongly related to marketing efforts.
In March, 即梦 went a step further, rising from ninth in February to sixth, but its competitor Kuaishou’s AI video tool 可灵 has not entered the top ten.
This may run counter to the perception of many people in the circle. More than one practitioner believes that 可灵’s text-to-video and image-to-video effects exceed those of the foreign Sora. In March of this year, Artificial Analysis, a world-renowned AI benchmark testing organization, released the latest global video generation model list. 可灵 1.6pro (high-quality mode) ranked first in the image-to-video track.
Some practitioners analyze that, on the one hand, AI video generation tools have a certain threshold, and naturally have fewer users than other tools. On the other hand, many users mainly use the Web version instead of the APP version, so when comparing the two, it is also necessary to refer to Web-side data.
The Xsignal list shows that 可灵 and 即梦’s Web-side rankings have been close to each other for the past three months, with 即梦 slightly ahead of 可灵. Andreessen Horowitz (a16z), a world-renowned investment fund and consulting company, released the 2025 Global Top 100 Generative AI Application Ranking, showing that 可灵’s web page had 20 monthly independent visitors, surpassing overseas well-known products such as Sora, Midjourney, and Runway, while 即梦 did not make the list.
Whether 可灵 or 即梦 has higher market recognition may have to wait until these applications have more users to draw a more accurate conclusion.
From the changes in rankings, it can be seen that the most competitive track is the chatbot track. In the first three months of this year, 夸克, 豆包, and DeepSeek have been firmly in the first tier of the domestic AI application list. 元宝 is the biggest variable. It did not appear on the list in January. In February, it surpassed 文小言 and Nano AI Search to rise to fifth, and in March, it surpassed Kimi to rank fourth. The remaining few are either standing still or declining in ranking month by month.
It should be pointed out that 元宝’s rapid climb is closely related to accessing DeepSeek. Although 夸克 also accesses DeepSeek,the main reason for its popularity is that it is rich in functions and convenient, bringing together AI search, AI dialogue, and different needs such as AI PPT, AI translation, and AI image generation.
Looking at the quarterly battle situation, three key veins are gradually becoming clear: First, multi-line layout has become a standard configuration for leading players. For example, ByteDance is betting on three different directions of AI applications at the same time: 豆包, 即梦, and 猫箱. MiniMax is involved in AI video and AI companions.
Second, the marketing war is intensifying.
Tencent 元宝 was able to rise in February, in addition to eating the DeepSeek dividend, it also used a large number of advertisements on and off the platform, and finally大力出奇迹. App Growing data shows that its estimated investment amount in the entire Q1 reached 1.7 billion, only tens of millions in January, and the investment amount directly reached 300 million in February, and it was even close to 1.4 billion in March (the investment amount is mainly calculated by combining the number of advertising creative ideas and list prices of the APP in the corresponding media, because most of 元宝’s investment is in Tencent channels, the actual cost may be lower than this amount). Under the big factories’ money-burning strategy, Kimi, which once invested heavily, gradually reduced its investment efforts, leading to a decline in its ranking.
App Growing data shows that the top ten AI applications with the most active MAU in China (DeepSeek and 夸克 do not have relevant data) also maintain high investment. Their total monthly investment amount from January to March reached 400 million, 500 million, and 1.6 billion, respectively. That is to say, the top popular AI applications in China spent 2.5 billion on advertising in one quarter.
Third, technological advantages and market performance are not completely positively correlated. 可灵 and 即梦’s recognition at the practitioner end is in the opposite state to the user-end monthly activity and downloads.
AI applications have not yet completely separated the winners and losers. In this arena, yesterday’s disruptors may become tomorrow’s defenders. Whether they can balance technological innovation and market insight tests every player.