Key Points
- Alibaba released 100+ open-source AI models called Qwen 2.5 for various sectors.
- The upgraded Qwen-Max 2.5 surpasses rivals like Meta’s LLaMA and OpenAI’s GPT-4 in reasoning and comprehension.
- A text-to-video tool allows users to create videos from simple prompts, similar to OpenAI’s Sora.
- Alibaba aims to challenge international cloud leaders like Amazon and Microsoft, hoping its AI advancements will boost cloud services growth.
Alibaba has taken a major step in the artificial intelligence (AI) race by releasing over 100 open-source AI models, known as Qwen 2.5, designed for various applications from automotive to gaming and scientific research. These models launched on Thursday, boast advanced math and coding capabilities and are part of Alibaba’s broader effort to compete with domestic and international tech giants.
The Chinese tech behemoth, headquartered in Hangzhou, aims to challenge rivals like Baidu and Huawei, as well as U.S. heavyweights Microsoft and OpenAI. Alibaba has been strategically expanding its AI offerings to attract developers and researchers globally by making these models open-source, allowing users to build generative AI applications without the high costs of training their systems.
Qwen 2.5 models are trained on large data sets, enabling them to understand user prompts and generate texts and images. The open-source nature of the models means that anyone—from academics to businesses—can access and utilize them, potentially increasing Alibaba’s influence in the AI space. Since launching its first version, Tongyi Qianwen (Qwen), last year, Alibaba has seen substantial interest, with its open-source models downloaded over 40 million times.
In addition to the open-source Qwen models, Alibaba upgraded its proprietary AI technology. The new flagship model, Qwen-Max 2.5, has been enhanced for improved reasoning and language comprehension, outperforming competitors like Meta’s LLaMA and OpenAI’s GPT-4 in several key areas. Unlike Qwen 2.5, Qwen-Max 2.5 is not open-source, but Alibaba offers it through its cloud computing division as a business service.
One of the most notable AI tools Alibaba introduced is a new text-to-video feature. Like OpenAI’s Sora, this tool allows users to input a prompt, and the AI generates a video based on that input. This innovation opens new possibilities for content creators and businesses looking to automate video production.
Eddie Wu, who took over as Alibaba’s CEO last year, emphasized the company’s focus on AI research and global infrastructure in a statement, noting that Alibaba is investing “with unprecedented intensity” in these areas. Wu has been actively working to reignite Alibaba’s growth as it faces challenges from increased competition and a slower consumer market in China.
While Alibaba is one of China’s largest cloud computing players, it still trails behind global giants like Amazon and Microsoft. With its latest AI offerings, the company hopes to attract more domestic and international customers, aiming to boost its cloud division, which saw promising growth signs in the June quarter after a sluggish period.
