--- title: "OpenAI's Late-Night Blockbuster Gets Mocked? PolyBuzz Tops the Chart, ByteDance Pulls Away, and Chinese AI Goes Global | November 2025 AI Top 100" author: "Unique Research" sourcePublication: "Unique Research Substack" originalPublishedAt: "2025-12-19T08:01:23+00:00" canonical: "https://ffcap.cn/en/research/src-20251219-02html" source: "https://uniqueresearch.substack.com/p/src-20251219-02html" language: "en" --- # OpenAI's Late-Night Blockbuster Gets Mocked? PolyBuzz Tops the Chart, ByteDance Pulls Away, and Chinese AI Goes Global | November 2025 AI Top 100 _Original · Unique Research · 2025-12-19_ _Historical edition: This is the complete textual article published on December 19, 2025, discussing November 2025 data. Product assessments, methodology, accuracy and partnership claims, and forecasts are those of the original report. The discussion places then-recent model releases alongside earlier November usage data; it does not establish that those releases caused November's results. The opening refers to 100 apps while the detailed app sections specify TOP30; both source scopes are retained. Cross-product active-user totals are sums, not deduplicated individual audiences. Images and image-only data are outside this text-only edition._ Following AI news over the past two days has produced a subtle sense of contrast. OpenAI released its new flagship image model, GPT Image 1.5, late at night. Its benchmark scores look strong and its editing is more stable, yet many users mocked it as a budget version of Nano Banana Pro; immediately after reading that discussion came ByteDance's release of Seedance 1.5 Pro, which generates videos with sound directly from text and images, synchronizes audio and visuals, maintains shot continuity, and represents a textbook multimodal power move. Focusing only on this news makes it easy to think the entire industry is debating whose images have finer detail and whose videos run more smoothly. Yet opening Unique Research's November data on Chinese AI products going global reveals another side: the products that truly capture overseas users' time compete less on spectacle than on solving several simple problems—editing photos, cutting videos, finding information, doing homework, and finding someone to talk to. Models are battling over multimodality in the sky while businesses slowly grow on the ground. In November, those two lines happened to intersect. Start with the Data: What Hundreds of Millions of Users Are Actually Using We tracked the top 100 Chinese AI Web products going global and the top 100 Chinese AI Apps going global. Consider only a few key figures: on the Web, the top 100 products collectively received about 440 million visits in one month, with combined monthly active users of roughly 170 million; among Apps, the top 100 generated approximately 80 million new downloads, while monthly active users approached 400 million. The top ten Web products by visits captured nearly half of all traffic; in overseas markets, only that small group of products truly captures users' attention. Foundation models upgrade their version numbers every day, yet ordinary users keep opening the same handful of rather ordinary-looking Apps and websites. On the Web, four categories—image generation, image editing, video generation, and video editing—collectively captured nearly half of all visits; among Apps, the highest downloads and monthly active users also center on the same cluster: photo editing, video editing, and image generation, supplemented by chat, learning, search, and companionship. From this perspective, the ground where GPT Image 1.5 and Seedance 1.5 Pro must take root is already visible in the table. PolyBuzz Fights Its Way Back to No. 1 If you looked only at the news, you might guess that the No. 1 Web product in November would be a major company's chatbot or a sensational video-model demo. The data gives a somewhat surprising answer: PolyBuzz has fought its way back to the top. This is an emotional-companionship product. In November, it recorded more than 30 million visits. Behind it came Alibaba's Qwen chatbot, the image-generation platform SeaArt AI, and various intelligent search and image-editing tools. The situation is intriguing: after going global, the most successful individual product from a company skilled at teaching children to solve problems was not an AI homework assistant, but a service that provides emotional soothing for adults. The same ranking includes emotional-companionship names such as CrushOn.AI, while Talkie, Saylo, Emochi, and other products hold stable positions on the App side. What does this show? On one hand, the story of AI as a tool has been told to exhaustion: helping you write copy, code, and emails. Everyone does it, and users are accustomed to it. On the other hand, when every platform can offer you a reasonably intelligent chat box, understanding emotions and knowing how to speak with you become differentiators. As computing power gradually becomes less scarce, emotional value becomes the new scarce resource. Products such as PolyBuzz sit precisely on this line. Rather than merely providing chat companionship, they move education products' accumulated understanding of people and interaction into emotional scenarios, retaining users through more delicate design. The App Side Becomes a Showcase for ByteDance's Ecosystem In the November ranking of Chinese AI Apps going global, adding up the monthly active users of ByteDance products produces a striking figure: approximately 140 million. Cici AI, Hypic, Gauth, Seekee, and related names together account for one-third of the entire ranking; by downloads, ByteDance also captured more than 40% of new installations. Viewed against the recent release of Seedance 1.5 Pro, the picture becomes even more vivid. Cici handles chat and serves as many people's gateway to Chinese foundation models; Hypic turns the small task of photo editing into a set of aesthetic tools; Gauth steadily turns photographing and solving homework problems into part of overseas students' daily routine; Seekee finds an appropriate position between search and assistant. Seedance 1.5 Pro advances audio-visual synthesis substantially, and these Apps already occupy the soil that it can directly empower. Chat no longer means only text; it can become a short video with narration at any time; photo editing is no longer limited to static images, but can become moving audio-visual content with one click; knowledge in education and search can also be packaged into video suited to distribution on social platforms. The model matters because it helps ByteDance turn underlying capabilities into a platform; these Apps run every day on overseas users' phones, gradually packaging audio-visual capabilities into specific small features. In the Video-Generation Battlefield, Users Care Only About What Works If one sector best represents the multimodal wave, it is unquestionably video generation. Kuaishou's KLING AI maintains visits on the order of ten million and appears steady, although it declined slightly month over month; MiniMax's Hailuo AI is quietly climbing, with both visits and monthly active users growing at double-digit rates. New faces have also appeared. One version update at MovieFlow, for example, multiplied both visits and active users several times over. Video users are actually not very loyal. They care little about which foundation-model company stands behind a product and spend little time reading technical descriptions. They look at only a few things: who generates more consistently, whose faces do not break, whose movements look more natural, and whose pricing offers better value. As soon as one product moves clearly ahead on these details, users surge toward it like a tide; if its iterations fall even half a step behind, the tide recedes equally quickly. In this environment, the release of a model such as Seedance 1.5 Pro essentially accelerates the reshuffling of the entire sector. Whoever can place integrated audio-visual capability into a genuinely usable video tool most quickly has a better chance of climbing several places in the next ranking. Growth Rankings Reveal Scenarios, Not Buzzwords For many people, the most attractive part of a ranking is its list of fastest-growing products. In November's overseas data, Livex grew fastest in Web visits. This human-like AI Agent helps B2C companies handle the entire customer journey, placing customer service, sales, and retention in the hands of AI employees who work around the clock, take no leave, and coordinate across online and offline channels. \[The source literally says “do not go to work for 24 hours,” an apparent wording error; the surrounding passage describes always-available AI staff.\] Once results stabilize and integration becomes smooth, the outcome appears directly in the data—visits and monthly active users rise almost in parallel. The same growth ranking includes video generator MovieFlow, agent platform MuleRun, speech-to-text tool Typeless, document assistant PopAi, and others. These apparently unrelated products are really doing the same thing: using AI to make one specific small task work sufficiently well. Products that truly break out rarely claim, “I can do everything.” They prefer to make one sentence clear: “I help you solve photographed homework problems,” “I help you edit videos,” or “I let you chat with a certain kind of character.” The more general the model, the narrower the application must be. Downloads Slow While Usage Deepens Another set of data may attract less attention than who ranks first, but it is critical for judging the trend. Among Chinese AI Apps going global in November, downloads declined for more than 60% of products, and the median overall download-growth rate had already turned negative. Yet among the same products, monthly active users were still rising for roughly two-thirds of Apps, and the median overall monthly active-user count increased slowly. Users no longer install several more Apps merely because they see a newly released foundation model. They are more cautious about choosing the few tools that remain on their phones, while products that have accumulated a user base raise usage frequency bit by bit through feature iterations, experience improvements, and localized operations. This closely resembles what is happening at the model layer. GPT Image 1.5 does not introduce a sensational new trick, but concentrates more on editing stability, speed, and workflow integration; Seedance 1.5 Pro also goes beyond making spectacular demos and improves indicators closely connected to actual production, including audio-visual synchronization, longer videos, and shot continuity. Models optimize for production while applications refine habits. Chinese AI products going global sit precisely at the intersection of these two changes. For those still standing on the shore choosing a boat, the ranking data may deserve more repeated reading than several model-launch events. It continually reminds us of a few simple questions: Exactly which users do you want to serve, and which minutes of their day are you trying to capture? Can you explain what you do in one sentence so that a user who understands nothing about AI can still grasp it? As models upgrade faster and faster, will your advantage lie in which model you use, or in understanding this one small task better than anyone else? The wave has already risen, and it will continue rising. What determines how far you travel, however, is ultimately not the wave but the boat you personally sail into it. The November data table on Chinese AI products going global helps us see at least one thing clearly. Most leading Chinese AI products going global no longer agonize over whether they are the most comprehensive. Instead, they quietly answer the same question: for one small task, is my product already useful enough that users will open it again tomorrow? Unique Research - An Authoritative Institution for Global AI Market Insights Details of Unique Research's AI Top 100 rankings for November 2025 are as follows: - Chinese AI Going Global WEB Ranking: TOP100 Chinese AI Web products going global, ranked by visits and monthly active users. - Chinese AI Going Global APP Ranking: TOP30 Chinese AI App products going global, ranked by downloads and monthly active users. Our Influence and Recognition: Unique Research's data and insights have been widely trusted and cited by leading investment institutions, academic organizations, and media outlets in China and abroad, becoming an important benchmark for their market analysis, investment decisions, and trend assessments. Our partners and citing organizations include, but are not limited to: Securities Firms and Investment Banks: Sinolink Securities, China Securities, Huayuan Securities, Haitong Securities, and others Authoritative Media: China Securities Journal, TMTPost, Jiemian News, and others We firmly believe that only data capable of withstanding validation by the industry's core participants has genuine long-term value. We believe in the power of open source and openness. Unique Research therefore makes the following three commitments: The AI Top 100 ranking will remain permanently open source and freely accessible, with the complete methodology and raw data supporting full reproduction. Data for the AI Top 100 ranking is strictly limited to independent third-party monitoring platforms. Data self-reported by companies is not accepted for evaluation. The AI Top 100 ranking system remains absolutely neutral and does not accept any commercial collaboration or payment intended to influence ranking results. How Should You Use the AI Top 100 Ranking? If You Are an Investor/Analyst: This ranking provides a clean data benchmark that can withstand reverse-engineering verification. It is a reliable foundation for building investment models, discovering under-the-radar projects, and conducting cross-validation. If You Are an AI Entrepreneur/Product Manager: You can not only identify competitors precisely, but also reproduce our methodology in depth to analyze your own product's market performance and optimize its iterations. If You Are a Researcher/Student: The fully open-source nature of this report makes it an excellent free resource for academic research and market analysis. We encourage innovative research based on this data. For the complete data behind the charts and rankings displayed in the AI Top 100, visit 100aiapps.cn, where all visible data can be copied and is freely available for personal study, research, and other noncommercial uses. Commercial use requires prior authorization from Unique Research. We welcome ranked teams, subject to compliant use, to apply their rankings in business plans, recruitment materials, and other contexts as authoritative evidence of their capabilities. Join the AGI Think Tank Open-Source Community The AI Top 100 is more than a ranking. It is an open-source community dedicated to advancing transparency in AI market data. We invite developers, data scientists, and industry analysts to: 1\. Reproduce and verify: visit our AGI Think Tank to obtain the complete data and methodology, validate our results, and propose improvements. 2\. Build together: contribute code and help us expand the data dimensions, including estimates of API call volumes and models for monitoring agents. 3\. Exchange insights: join the AGI Think Tank discussion group and share distinctive conclusions derived from Unique Research data with peers. One person's insight is limited; the wisdom of a group is boundless. Let us define the standards for measuring the AI market together. 01\. Chinese AI WEB Products Going Global Ranking 1.1 Chinese AI WEB TOP 100 by Monthly Visits Based on its independent research into the AI industry and continuous tracking of thousands of AI companies and nearly ten thousand AI products worldwide, Unique Research ranks AI WEB products by their cumulative visits for the month. It has compiled the following ranking of the TOP 100 AI WEB products created by Chinese or ethnically Chinese entrepreneurs for overseas markets, ordered by visits: \*For the complete data, visit 100aiapps.cn, where all visible data can be copied \*For the complete data, visit 100aiapps.cn, where all visible data can be copied 1.2 Chinese AI WEB TOP 100 by Monthly Active Users Based on its independent research into the AI industry and continuous tracking of thousands of AI companies and nearly ten thousand AI products worldwide, Unique Research ranks AI WEB products by their active users for the month. It has compiled the following ranking of the TOP 100 AI WEB products created by Chinese or ethnically Chinese entrepreneurs for overseas markets, ordered by active users: \*For the complete data, visit 100aiapps.cn, where all visible data can be copied \*For the complete data, visit 100aiapps.cn, where all visible data can be copied 02\. Chinese AI APP Products Going Global Ranking 2.1 Chinese AI APP TOP 30 By Monthly Downloads Based on its independent research into the AI industry and continuous tracking of thousands of AI companies and nearly ten thousand AI products worldwide, Unique Research ranks AI APP products by their cumulative downloads for the month. It has compiled the following ranking of the TOP 30 AI APP products created by Chinese or ethnically Chinese entrepreneurs for overseas markets, ordered by downloads: \*For the complete data, visit 100aiapps.cn, where all visible data can be copied \*For the complete data, visit 100aiapps.cn, where all visible data can be copied 2.2 Chinese AI APP TOP 30 By Monthly Active Users Based on its independent research into the AI industry and continuous tracking of thousands of AI companies and nearly ten thousand AI products worldwide, Unique Research ranks AI APP products by their active users for the month. It has compiled the following ranking of the TOP 30 AI APP products created by Chinese or ethnically Chinese entrepreneurs for overseas markets, ordered by active users: \*For the complete data, visit 100aiapps.cn, where all visible data can be copied \*For the complete data, visit 100aiapps.cn, where all visible data can be copied 03\. Research Statement 3.1 Data Notes The AI product revenue data in this report, article, ranking, or chart is based on real-time tracking of traffic redirected from AI product websites to payment gateways such as Stripe, combined with time-series analysis. This process examines historical visits, fluctuations in monthly active users, and changes in pricing strategy in depth to construct a dynamic revenue-estimation model. The model's core parameters are determined through a three-part validation system comprising paid-conversion-funnel monitoring, reverse calibration using company-disclosed data, and comparison with industry benchmark conversion rates; they are also promptly compared with and corrected against corporate financial reports, financing announcements, and disclosures from authoritative media. Sample validation shows that the mean absolute error rate of predictions remains within ±10%, meeting industry-grade research standards. The AI product traffic data in this report, article, ranking, or chart is based on the integration of multiple data sources and processing by Unique Research's intelligent algorithms. Monitoring is strictly limited to direct user visits through official applications and websites. Specifically, the data covers only behavior in which users directly access AI products through websites and native applications. It excludes traffic generated through browser plugins/extensions, desktop client software, WeChat/Alipay mini-program ecosystems, embedded services on third-party platforms such as Discord, local deployment of open-source models, API calls, and other indirect-access scenarios. The data focuses on monitoring direct use of AI products on core user endpoints, with the aim of objectively reflecting direct use on mainstream user devices. Given the complexity and dynamic nature of data collection and processing, as well as the market environment's continual change, the displayed data may contain a certain degree of error and omission. It should therefore be regarded as a reference for research and analysis, intended to provide users with a window into the AI product market, rather than as a definitive basis for specific investment strategies or consulting advice. Statement on Expanded Data Coverage: We are actively developing and testing monitoring models for the following emerging fields, which are expected to be included gradually in future quarterly reports: AI agent (Agents) ecosystems AI hardware-device sales and activity Estimates of large-model API call volume Stay tuned. 3.2 Definitions • Geographic Dimension Overseas AI Products: Products founded by entrepreneurs or teams not of Chinese heritage and launched primarily for global markets, including but not limited to their domestic markets. Domestic AI Products: Products founded by entrepreneurs or teams of Chinese nationality and launched primarily for China's domestic market. Chinese AI Products Going Global: Products founded by entrepreneurs or teams of Chinese heritage but targeted primarily at overseas markets outside mainland China. • Functional Dimension AI-Native Applications: Applications in which artificial intelligence technologies and algorithms are deeply integrated from the outset of product design, and whose core value, business processes, or user experience depend entirely on AI. These applications do not merely use AI to optimize or enhance existing functions; AI is a core component, without which the applications would cease to exist or lose their core value. AI-Enabled Applications: Applications that integrate or embed AI technology into an existing business logic and application framework to enhance current functions or provide entirely new ones. These applications may already have mature business models and market positioning, but by introducing AI, they can significantly improve efficiency, improve user experience, or create new value. AI-Ecosystem Applications: Platform applications that support, connect, or facilitate communication among AI technology researchers, application developers, service providers, end users, and the broader AI ecosystem. They are not limited to the technical layer, but also span markets, communities, resource sharing, and other dimensions, with the aim of promoting the adoption, innovation, and collaborative development of AI. 3.3 Metrics • WEB Data Metrics Visits: The total number of visits to a website during a specified period, such as one month, used to measure the scale of website traffic. Page views during a continuous period of activity count as the same session, while renewed activity after an interval of more than 30 minutes or at the beginning of a new day is counted as a new visit, thereby providing a comprehensive and precise measure of traffic and visitor activity. Unique Visitors: The number of unique IPs visiting the target website during a specified period, such as one month. If someone visits a website on multiple days during a month, that person is counted as only one unique visitor. Subscription Revenue (labelled ARR in the source): The total revenue a website earns from subscription services provided to users over a specified period, such as one year. It excludes advertising revenue, transaction commissions, and one-off revenue such as professional services. This retains the source's period-revenue definition rather than substituting the usual annualized recurring-revenue metric. • APP Data Metrics Downloads: The number of times an App is downloaded during a specified period, such as one month. It is generally measured by downloads of a single App from application stores such as the App Store and Google Play, and is an important indicator of an App's popularity and user appeal. It can also reflect the effectiveness of App promotion and marketing activities. Active Users: The number of unique users who engage in at least one activity during a specified period, such as one month. Active users measure user participation and activity during a specific period and are generally used to evaluate an App's user scale and engagement. In-App Purchase Revenue (IAP): Revenue generated by purchases that users make within an App to obtain virtual goods, additional services, premium features, and similar offerings. It excludes advertising revenue, direct user payments such as tips, and revenue from third-party Android app stores. 3.4 Free-Use Statement This report is published by Unique Research and is copyrighted by Unique Research. Any Chinese-language republication or citation must identify the report's source; overseas institutions seeking to republish or cite it should contact us in advance for authorization. When citing data, use the attribution: \[Data source: Unique Research\]. When directly citing charts, use: \[Data source: Unique Research; charts produced by Unique Research\]. This report is an independent, original analysis produced by Unique Research as a third-party institution. Its content does not represent the position of any company and does not constitute investment advice to any person. Investors must therefore note that Unique Research, its employees, and its affiliates bear no responsibility for investment decisions made on this basis. Where permitted by law, Unique Research and its affiliates may hold equity in companies mentioned in this report, or may provide or seek to provide them with financing, financial-advisory, or related services. Employees may serve as directors of companies mentioned in the report. Appendix: Summary of Quarterly Trend Insights Based on the data in this edition, the Unique Research analyst team identified the following core trends: First, visual content remains the most reliable gateway. Across both Web and App products, image- and video-related offerings occupy the core of traffic and active users. Photo editing, video editing, and video generation are high-frequency, essential needs and are naturally suited to expansion across languages and cultures. They will remain an unavoidable main line when observing Chinese AI products going global. Second, emotion and companionship are moving from “gimmicks” to “genuine categories.” PolyBuzz, CrushOn.AI, Saylo, and Emochi share one characteristic: they treat AI as a character capable of long-term interaction rather than a one-off question-and-answer tool. The data already shows stable performance for emotional-companionship products in visits, downloads, and monthly active users. In the future, this area will test content design and long-term operations, not merely model capability. Third, Chinese Agent and utility products going global have entered the validation stage. Growth at Livex, MuleRun, Typeless, and various search and writing tools shows that overseas users are accepting agents that “help me solve a specific problem.” Compared with a unified “super Agent,” these products are more grounded: one handles customer service, another search, and another speech transcription. They focus on one or two steps in a business chain rather than taking over the entire process. Fourth, major-company product portfolios and startup teams will coexist for the long term. Major companies' advantages in distribution, computing power, model capabilities, and cross-product coordination are difficult to dislodge, so ByteDance products will continue to occupy 30% to 40% of the App pool discussed in this report. Yet each new interaction form and user segment still offers many opportunities for small products to go deep. Those opportunities may not appear in the global download Top5, but they may well be hidden in the month-over-month growth Top10 list. --- Original publication: https://uniqueresearch.substack.com/p/src-20251219-02html On-site reading page: https://ffcap.cn/en/research/src-20251219-02html