Alibaba has released over 100 new open-source artificial intelligence (AI) models, significantly expanding its AI offerings in a bid to compete with major players in both China and the U.S. The new models, branded as Qwen 2.5, are designed for diverse sectors such as automotive, gaming, and scientific research. They bring enhanced math and coding capabilities, according to Alibaba. The company’s flagship proprietary model, Qwen-Max, also received an upgrade, positioning it above competitors like Meta’s Llama and OpenAI’s GPT-4 in areas such as reasoning and language comprehension.
Alibaba’s strategic move to open-source these models allows users globally—academics, companies, and researchers—to integrate the technology into their own generative AI applications without needing to build systems from scratch. This move is aimed at fostering widespread adoption of its AI solutions. Since launching its first AI model, Tongyi Qianwen (Qwen), last year, Alibaba has reported 40 million downloads of its open-source models.
Apart from the models, Alibaba has introduced a new text-to-video generation tool, allowing users to generate videos from prompts, a technology similar to OpenAI’s Sora. The company views this as part of its broader plan to strengthen its AI capabilities and boost growth within its cloud division. Eddie Wu, Alibaba’s CEO, emphasized the company’s commitment to AI development and global infrastructure building as key elements in its future strategy.
While Alibaba remains a dominant player in China’s cloud computing industry, it lags behind Amazon and Microsoft internationally. The company hopes its expanding AI portfolio will attract both domestic and international customers, strengthening its cloud services business, which showed early signs of recovery in the latest quarter despite ongoing challenges from competitors and a sluggish consumer market in China.
