Alibaba’s Qwen surpasses 10 million downloads, challenges AI giants
Alibaba Group Holding announced on Monday that its new artificial intelligence assistant app, Qwen, achieved over 10 million downloads in just the first week of its public beta launch, with the South China Morning Post reporting that Qwen’s early performance surpasses that of OpenAI’s ChatGPT and DeepSeek during their initial launch phases. Launched last week, Qwen is a significant step in Alibaba’s initiative to develop a formidable AI assistant capable of competing with global leaders such as Google’s Gemini and ChatGPT, both inaccessible in mainland China. Alibaba’s stock, listed in Hong Kong, concluded the trading day with a notable increase of 4.67 per cent, reaching HK$154.50 on Monday, and the company is set to unveil its quarterly results on Tuesday. In a statement, Alibaba remarked that Qwen represented “Alibaba’s most significant step yet into the consumer AI market, aiming to transform its cutting-edge foundational AI model capabilities into real-life applications and tools.”
Qwen operates on Alibaba Cloud’s open-source AI model family, also referred to as Qwen, and is crafted to assist users in managing both personal and professional tasks, with capabilities including in-depth research, producing images, and crafting presentation slides. Alibaba Cloud, the cloud computing and AI division of the Hangzhou-based company, stated that Qwen would ultimately evolve into an all-encompassing “everything app”, currently accessible on Apple’s iOS and Android platforms. The rapid uptake of Qwen in its inaugural week bolsters Alibaba’s assurance in expanding its reach to a broader global audience, with the company intending to introduce an international version that will encompass maps, food delivery, travel bookings, office productivity tools, e-commerce, education, and health services. A report indicates that new versions of Qwen will be rolled out swiftly.
According to the reports, Alibaba’s advanced AI model Qwen3-Max outperformed five major global AI systems in a live cryptocurrency trading experiment conducted by the US research firm Nof1. In the initial phase of the “Alpha Arena” competition, Qwen3-Max achieved a return of 22.32 percent on a starting investment of $10,000 within two weeks, with only two of the six models generating a profit. China’s DeepSeek V3.1 Chat achieved second place with a 4.89 per cent return, while four prominent US-based models—OpenAI, Anthropic, Google DeepMind, and Elon Musk’s xAI—reported significant losses, with OpenAI’s GPT-5 performing the worst, losing 62.66 per cent of its investment. Nof1 noted that the initial findings “may be the result of luck” and stated that future rounds would incorporate “more statistical rigour”.
The Alpha Arena experiment restricts AI tools to quantitative market data exclusively, omitting news or external information, prompting examination of how these systems might function in real investment situations. Interestingly, Qwen3-Max was the only AI model in the test lacking reasoning capabilities, indicating it did not engage in step-by-step thinking before making trading decisions. The results, while striking, raise questions about model design, risk handling, and market adaptation in high-frequency or algorithmic trading scenarios. As Alibaba continues accelerating Qwen’s global rollout and AI ambitions, the early traction and experimental results have amplified industry attention on China’s emerging AI competitors and their rapidly evolving capabilities.
