Alibaba creates a rival to DeepSeek, causing shares to soar.

On Thursday, the Chinese internet giant Alibaba released its most recent AI reasoning model, claiming that it outperformed competing models from OpenAI and startup DeepSeek.

Alibaba’s Hong Kong-listed shares ended the day 8% higher as a result of the news, which also helped the Hang Seng’s China Enterprises Index increase.

Alibaba’s latest AI model was released one day after another business introduced Manus, a “general AI agent.” The software can perform intricate, multi-step operations, including screening resumes and building a website, according to a video on the Manus website. Manus is the product of the Chinese business Monica, according to Reuters.

According to the video, the AI agent is more sophisticated than a chatbot as it produces measurable outcomes in addition to ideas, such as a report that suggests properties to purchase based on predetermined standards.

In an online statement, Alibaba praised its new model, QwQ-32B, for achieving “exceptional performance, nearly completely outperforming OpenAI-o1-mini and competing with the most powerful open-source reasoning model, DeepSeek-R1.” Last year, the American company’s cost-effective reasoning model, OpenAI-o1-mini, was unveiled.

In the announcement, Alibaba also noted that the model had made a “qualitative leap in mathematics, coding, and general capabilities, with overall performance on par with DeepSeek R1.”

According to the business, DeepSeek’s R1 model includes 671 billion parameters, whereas its model has 32 billion. A model with fewer parameters is likely smaller and easier to train.

In January, DeepSeek shocked the world with its R1 reasoning model, which it claimed was significantly less expensive to train than its well-known Western competitors. As the US-China tech rivalry has heated up, its success has bolstered trust among international investors in Chinese companies’ capacity for innovation. Since January, the Hang Seng China Enterprises Index has increased by more than 30%.

After OpenAI released its game-changing AI reasoning model in 2023, Alibaba, the owner of the Chinese e-commerce sites Taobao and Tmall, initially introduced its ChatGPT-equivalent service, Tongyi Qianwen.

Alibaba stated that its Qwen 2.5 Max model, which was published in January, outperformed DeepSeek’s highly regarded V3 model, which had been out for only a few weeks.

Alibaba promised last week to invest at least 380 billion yuan ($52.4 billion) in its cloud computing and artificial intelligence infrastructure over the following three years. It said that the sum was more than what it had spent on those fields during the previous ten years.

Chinese officials promised to assist “emerging industries and industries of the future” on Wednesday, including boosting investments in quantum technology, humanoid robots, and artificial intelligence.