The Alibaba Qwen3 APP crashed on its first day! Qwen3 supports 600 million downloads, ranking first globally.

On November 17, Alibaba announced that its personal AI assistant Qianwen APP has officially opened for public testing. The Qianwen APP is based on the world's top-performing Open Source model Qwen3, positioned as a personal AI assistant that can both “converse” and “get things done.” On the first day of public testing, traffic surged, causing some services to experience interruptions, leading to “Alibaba Qianwen crashed” quickly trending on Weibo. The official response stated that the system is in good condition and welcomed users to participate.

k Questions APP saw explosive traffic on its first day, users experienced congestion

Alibaba officially launched the open beta of the Qwen APP on November 17, making it available to users for free. The Qwen APP is positioned as a personal AI assistant, built on the globally top-performing Open Source model Qwen3, capable of both “conversation” and “task execution.” Alibaba plans to link life scenarios such as maps, takeout, ticket booking, office work, learning, shopping, and health to the Qwen APP, allowing users to complete daily tasks through a single application. The public beta version has been launched on multiple app stores, and the web and PC versions are also available.

On the first day of the public test, the traffic of the Qianwen APP surged, leading to some service interruptions. Many users reported that they could not receive a response after entering commands, with the application prompting “entrance blockage.” This situation is not uncommon during the debut of technology products in China, but the peak traffic of the Qianwen APP indicates high market expectations for Alibaba's AI products. The topic “Alibaba Qianwen crashed” quickly rose to trending on Weibo, with discussion volume exceeding millions within a few hours.

Alibaba's officials did not evade the issue but responded in a relaxed tone, saying: “I’m doing well, in good condition, feel free to ask.” This public relations strategy transforms potential negative public sentiment into positive product popularity, demonstrating Alibaba's maturity in operating in the C-end market. The officials emphasized that the overall system is operating normally, and some congestion is a normal phenomenon during peak traffic periods. The technical team is rapidly expanding capacity to meet the continuously growing user demand.

The design of Qianwen APP is extremely minimalist, with the networking search function enabled by default. This design aims to reduce the risk of “hallucinations” in large models. “Hallucinations” refer to instances where AI models generate seemingly reasonable but actually incorrect content when lacking accurate information. By enabling online search by default, Qianwen APP can instantly retrieve the latest information, ensuring the accuracy and timeliness of its responses. Additionally, the application integrates features such as image generation, AI photo editing, video calls, and real-time translation, providing a one-stop AI service experience.

Qwen3 Global Performance First 600 Million Downloads Surpassing Llama

The Qianwen APP is built on the Qwen series of large models. Since Qwen was fully Open Source in 2023, it has outperformed international open-source models like Llama and Deepseek, with global downloads exceeding 600 million. This figure is extremely rare in the field of open-source AI models, demonstrating the high recognition of Qwen3 within the global developer community. The Open Source strategy allows global developers to conduct secondary development and optimization based on Qwen3, forming a vast ecosystem.

Airbnb co-founder Brian Chesky stated that the company heavily relies on Qwen, which he believes is faster and better than OpenAI models. This is a direct endorsement from a top tech company in Silicon Valley, which carries significant persuasive power. As the world's largest home rental booking platform, Airbnb's AI needs span multiple areas including customer service, content review, and personalized recommendations. Choosing Qwen3 over OpenAI models means that in practical business applications, Qwen3 demonstrates competitive advantages in performance, cost, and deployment flexibility.

NVIDIA founder Jensen Huang also pointed out that Qwen occupies a significant share in the global Open Source model market and is expanding. As a leading figure in the AI hardware field, his viewpoint serves as a barometer for the entire industry. NVIDIA's collaboration with Alibaba is also deepening, with both parties launching flagship cooperation in the PhysicalAI field to provide enterprise users with full-platform Alibaba Cloud services.

Three Core Advantages of Qwen3

Global Performance Number One: In the LMArena text rankings, the Qwen3-Max-Instruct preview version ranks third, surpassing GPT-5-Chat.

Open Source Ecosystem is Strong: 600 million global downloads, the developer community's activity level far exceeds that of other open source models.

Commercial validation is sufficient: The actual deployment by top tech companies like Airbnb proves its commercial viability.

Alibaba released the Tongyi model Qwen3-Max and the next-generation foundational model architecture Qwen3-Next at the Yunqi Conference on September 24. The preview version of Qwen3-Max-Instruct ranks third on the LMArena text leaderboard, surpassing GPT-5-Chat. This ranking is not a promotion from Alibaba itself, but comes from a third-party authoritative evaluation platform, ensuring objectivity and fairness. Surpassing GPT-5-Chat means that Qwen3 has reached or even exceeded the level of OpenAI's latest model in specific tasks.

$380 billion investment in AI infrastructure to benchmark OpenAI

Alibaba announced earlier this year that it will invest 380 billion yuan in AI infrastructure and plans to increase its investment further. This investment scale ranks among the top in Chinese tech companies, demonstrating that Alibaba sees AI as the most core strategic direction for the next decade. The 380 billion yuan investment will be used in various aspects such as computing power construction, model research and development, talent introduction, and ecosystem cultivation.

Computing power is the foundation of AI development. Alibaba Cloud, as the largest cloud service provider in China, has a global network of data centers and strong computing power resources. Investing 380 billion yuan means that Alibaba will significantly expand the scale of its GPU clusters, particularly by procuring NVIDIA's latest H100 and H200 series GPUs. The flagship collaboration with NVIDIA in the PhysicalAI domain ensures that Alibaba can prioritize access to the most advanced AI hardware.

The Qianwen APP is led by Alibaba's Intelligent Information Business Group, with participation from multiple departments including Alibaba Cloud, Tongyi Laboratory, Taotian, and Gaode. This cross-departmental collaboration model is a key feature of Alibaba's AI strategy. Tongyi Laboratory is responsible for core algorithm development, Alibaba Cloud provides computing power and cloud service support, while Taotian and Gaode contribute practical application data from e-commerce and mapping scenarios. This integrated model of production, education, and research allows Qwen3 to not only lead in theoretical performance but also be fully validated in real business scenarios.

Alibaba stated that the public beta version is a basic version, and in the future, it will gradually add Agentic AI features, especially in the shopping experience on e-commerce platforms like Taobao. Agentic AI refers to AI systems that can autonomously complete complex tasks, rather than just answering questions. For example, users can tell the Qianwen APP, “Help me find a pair of running shoes suitable for running, budget 500 yuan,” and the AI will not only recommend products but also directly perform price comparisons and place orders on Taobao. This experience will fundamentally change the interaction model of e-commerce, shifting from “people looking for goods” to “AI looking for goods.”

Taobao, as the largest e-commerce platform in China, has an enormous amount of product data and user behavior data. Deep integration of Qwen3 with Taobao means that AI can learn each user's shopping preferences, price sensitivity, and brand inclination, providing highly personalized shopping recommendations. This capability is difficult for OpenAI's ChatGPT or other general AI assistants to match, as they lack deep integration with e-commerce platforms.

From a strategic layout perspective, Alibaba is building the Qianwen APP into a super entrance that connects users with various services of the Alibaba ecosystem. The integration of life scenarios such as maps, takeout, ticket booking, office work, learning, shopping, and health means that users do not need to switch between multiple apps and can complete all daily tasks through natural language dialogue. This vision of a “one-stop AI assistant” competes with Apple's Siri and Microsoft's Copilot, but Alibaba has richer ecological resources and deeper scene integration capabilities.