Qwen's 166M MAU Surge and Kimi's Fall: How China's AI Reshuffling in Q1 2026 Changes Your Brand Visibility Priorities

Six months ago, Kimi was China's hottest AI chatbot. Investors were pouring in, user counts were climbing, and brand marketers were told to pay attention to Moon Shot AI as the rising force in Chinese AI search. Today, Kimi has tumbled out of the consumer top tier — its monthly active users collapsed from a peak of 36 million in late 2024 to somewhere between 10 and 15 million by early 2026. Meanwhile, Alibaba quietly unified its entire AI ecosystem under a single brand name and surged to 166 million MAU. And Doubao, ByteDance's AI assistant, hit 345 million MAU in Q1 2026 — a number that reframes the entire competitive landscape.

For international brands managing China market visibility, this is not background noise. The AI models your target customers are using to research products, compare brands, and make purchasing decisions have shifted dramatically. If your GEO strategy is still calibrated to the China AI market of mid-2025, it is already misaligned.

This article unpacks what happened, why it matters for brand citations, and how to reprioritize your China AI visibility efforts accordingly.

What the Q1 2026 Numbers Actually Say

QuestMobile's Q1 2026 data, released in April, revealed the starkest concentration the Chinese AI assistant market has seen since the initial consumer wave of 2023. The top three platforms by monthly active users were:

These three platforms together account for the overwhelming majority of Chinese consumer AI usage. Doubao and DeepSeek alone command roughly 88% of China AI assistant monthly active reach, according to industry reporting from NBD.

Notably absent from this list: Kimi, Tencent Yuanbao, and Baidu's Wenxin, all of which had been cited as major players in 2024 brand visibility discussions. Kimi and Yuanbao both lost significant ground. Kimi's fall from second place to seventh place (by weekly actives, as of late 2025) is particularly stark — and has direct implications for any brand that optimized its GEO strategy around Kimi's citation behavior.

How Kimi Lost Its Early Lead — and Why It Matters for Brands

Kimi's initial rise was driven by its long-context capabilities and early mover advantage in conversational AI in China. In late 2023 and through most of 2024, Kimi was genuinely differentiated: it could process far longer documents than competing models, making it popular for research-heavy tasks. Brand marketers were right to take it seriously.

The problem was that long-context advantage became commoditized almost immediately. DeepSeek, Qwen, and eventually Doubao all extended their context windows while also delivering better raw performance and, crucially, massive distribution advantages. Doubao is embedded natively across ByteDance's ecosystem — Douyin, Toutiao, and the Feishu workplace suite. Qwen is integrated with Alibaba's Taobao, Tmall, and Alibaba Cloud. Kimi had no comparable distribution.

By Q3 2025, Kimi's MAU had dropped from 36 million to under 10 million, a decline of more than 70%. By December 2025, it ranked seventh in weekly active users with only 4.5 million. Tiger Brokers' research summarized the situation bluntly: "Money burned, users fled: once-glorious Kimi falls to second tier after just one year."

The brand visibility implication is significant. If your brand was cited prominently by Kimi in 2024 or early 2025, that visibility was reaching a much larger audience than it reaches today. A brand that optimized content and citation pathways specifically for Kimi's training signals and citation preferences now needs to reassess: is that optimization effort now serving 4.5 million weekly users instead of the 30+ million it once did?

Alibaba's Qwen Rebrand: Why 166M MAU Changes the Brand Visibility Calculus

On February 5, 2026, Alibaba announced a sweeping brand unification: all AI technologies, consumer products, and research would now operate under a single name — Qwen. Tongyi Qianwen, previously the flagship consumer app, was absorbed. All large language models became "Qwen models." The Tongyi Lab retained its name as the research entity, but the consumer face became entirely Qwen.

This matters for brands for three reasons.

First, the scale is now undeniable. With 166 million MAU, Qwen is not a niche research tool or a developer API. It is the second-largest AI assistant platform in China, actively used by consumers making purchase decisions, researching travel options, comparing product categories, and asking for brand recommendations. A platform at this scale has more citation influence than virtually any single media publication in China.

Second, Alibaba's ecosystem integration creates unique brand signal pathways. Qwen's recommendations are not trained in isolation. They are informed by data signals from the Alibaba ecosystem — which includes Taobao (China's largest C2C marketplace), Tmall (the dominant B2C platform), Alibaba Cloud, and international commerce platforms. A brand with strong presence, reviews, and transactional history on Tmall is building citation signals not just for traditional search but for Qwen's recommendation layer.

For international luxury brands, consumer goods companies, and fast-moving consumer goods players who have already invested in Tmall flagship stores: your Tmall presence is now a direct input into China's second-largest AI recommendation platform. This is not a coincidence — it is an architecture decision Alibaba made deliberately.

Third, the technical trajectory favors Qwen for brand-relevant queries. According to Digital Applied's Q2 2026 landscape report, Qwen 3.6 Plus holds 13.9% of OpenRouter weekly token share globally, processes 2.77 trillion weekly tokens, and handles multi-language brand queries with broad multilingual coverage. Qwen's always-on chain-of-thought reasoning means it tends to give more elaborated, source-cited answers to product and brand questions — exactly the format in which brand mentions either appear prominently or get filtered out.

Doubao at 345M: The Non-Negotiable Priority

If Qwen's rise is the most strategically interesting development, Doubao's scale is the most strategically non-negotiable fact.

At 345 million monthly active users, Doubao is not one channel among many. It is the channel. For any brand targeting Chinese consumers through AI visibility, failing to appear in Doubao's recommendations is equivalent to being invisible on Baidu's first page in 2015 — it means missing the dominant discovery surface entirely.

Doubao's citation behavior differs from Kimi's in important ways. Where Kimi historically favored long-form, research-oriented content and academic-adjacent sources, Doubao's consumer positioning means its recommendations reflect broader media coverage, social platform signals (particularly from Douyin and Xiaohongshu), and mainstream brand recognition. Brands with strong Chinese social media presences — active accounts, KOL partnerships, and consistent Douyin video presence — tend to perform better in Doubao's recommendation layer.

This is confirmed by the data from hubGEO's brand scoring methodology: brands that score highest across China's six major AI models share a common characteristic beyond just brand size. They have maintained consistent Chinese-language content ecosystems across multiple platforms, creating redundant citation pathways that any major AI model can draw from.

The New Brand Visibility Priority Stack

Based on Q1 2026 MAU data, here is how international brands should now weight their China AI visibility investment:

This is a meaningful change from the priority stack that made sense 12 months ago, when Kimi's 30M+ MAU placed it closer to P2 priority. The market has reshuffled, and brand visibility budgets and content strategies should reshuffle accordingly.

One important caveat: MAU is a proxy for reach, not a measure of decision-making influence. A smaller platform serving professionals or high-income consumers in specific categories may still warrant investment even at lower MAU if your brand's target segment concentrates there. Kimi's remaining user base, for instance, skews toward researchers, students, and knowledge workers — a relevant audience for B2B brands and certain premium consumer categories.

Takeaway for Brand Marketers

China's AI assistant market has structurally consolidated around three platforms — Doubao, Qwen, and DeepSeek — while earlier leaders like Kimi and Yuanbao have retreated to niche positions. This consolidation simplifies the prioritization problem while raising the stakes: getting visibility on the top three platforms now matters more than spreading attention evenly across six or seven.

Three actions to take now:

First, audit your Tmall and Taobao brand presence. If you have a flagship store, ensure product descriptions, brand story pages, and review responses are in accurate, natural Chinese. These are now direct inputs into Qwen's brand knowledge layer.

Second, evaluate your Douyin content consistency. Doubao's training signals draw heavily from ByteDance's own ecosystem. Brands that treat Douyin as optional are inadvertently deprioritizing themselves in China's largest AI platform.

Third, stop over-indexing on Kimi optimization. If your team spent 2024 tailoring content specifically for Kimi's long-context preference or citation style, that effort needs to be redirected toward platforms with an order of magnitude more monthly reach.

The market reshuffled fast. Your strategy should too.

Related: Track how your brand scores across all six China AI models →

Data sources: QuestMobile Q1 2026 AI Application Value Report (via Sina Finance, April 2026); Digital Applied Chinese AI Models Q2 2026 Landscape Report (April 2026); Tiger Brokers Kimi market analysis; hubGEO brand scoring methodology.