Hong Kong: Alibaba, a Chinese tech company, said on Thursday that its latest AI reasoning model outperformed competitors OpenAI and startup DeepSeek. The model was unveiled by Alibaba.
With the news breaking, Alibaba’s Hong Kong-listed shares closed the day up 8% and contributed to the China Enterprises Index’s surge on the Hang Seng.
Just one day after another business unveiled their “general AI agent” named Manus, Alibaba released their latest AI model. According to a video on the Manus website, the program can perform complicated, multi-step operations including website creation and resume screening. Reuters reports that the Chinese company Monica is the one responsible for creating Manus.
According to the video, the AI agent goes beyond being a chatbot and actually accomplishes things like creating a report with property recommendations based on user-specified criteria, rather than just coming up with ideas.
“Exceptional performance, almost entirely surpassing OpenAI-o1-mini and rivaling the strongest open-source reasoning model, DeepSeek-R1.” That was the claim made by Alibaba in an online statement praising their new model, QwQ-32B. The American company’s cost-efficient reasoning model, OpenAI-o1-mini, was introduced last year.
This model has made a “qualitative leap in mathematics, coding, and general capabilities, with overall performance on par with DeepSeek R1,” according to Alibaba’s announcement.
The business boasted that its model has 32 billion parameters, far more than DeepSeek’s R1 (671 billion). It is easier and faster to train a model with fewer parameters.
In January, DeepSeek shocked everyone with its R1 reasoning model, which it claimed to have trained for significantly less money than its well-established Western competitors. Despite the escalating US-China tech rivalry, its success has bolstered faith among international investors in the innovative capabilities of Chinese enterprises. There has been a 30% increase since January in the Hang Seng China Enterprises Index.
In 2023, after OpenAI’s launch of its groundbreaking AI reasoning model, Alibaba—owner of Taobao and Tmall—introduced Tongyi Qianwen, a service that was similar to ChatGPT.
According to Alibaba, their Qwen 2.5 Max model outperformed DeepSeek’s much-lauded V3 model, which had been released only a few weeks prior, in January.
Over the next three years, Alibaba plans to invest a minimum of 380 billion yuan ($52.4 billion) in its cloud computing and artificial intelligence infrastructure, according to a vow made last week. It claimed the sum was more than what it had spent in certain sectors throughout the previous decade.
Chinese officials increased financing for quantum technology, artificial intelligence, and humanoid robots as part of their commitment to “emerging industries and industries of the future” on Wednesday.