Alibaba doubles down on LLM race

Alibaba Cloud, the cloud computing unit of Chinese tech heavyweight Alibaba Group Holding Ltd, unveiled on Thursday two open-source artificial intelligence-powered large language models, a move industry experts said will bolster the technological advancement and industrial application of LLMs.

The company said it will open-source two LLMs called Qwen-7B and Qwen-7B-Chat, with each model boasting 7 billion parameters, which could be put into commercial use. This marks the first time that a big Chinese tech company has open-sourced its LLM.

Alibaba Cloud emphasized that the open-source LLM will help enterprises simplify the process of model training and deployment, as they just need to fine-tune the models and build their own high-quality AI models effectively and cost-efficiently.

"By open-sourcing our proprietary large language models, we aim to promote inclusive technologies and enable more developers and small and medium-sized enterprises to reap the benefits of generative AI," said Zhou Jingren, chief technology officer of Alibaba Cloud Intelligence.

For commercial use, the models will be free for use by companies with fewer than 100 million monthly active users. Companies with more users can request a license from Alibaba Cloud, the company said. The move comes after US tech company Meta rolled out a similar open-source model named Llama 2 last month.

Alibaba unveiled its LLM called Tongyi Qianwen in April, joining the chatbot race to offer a potential rival to ChatGPT. The launch of Qwen-7B and Qwen-7B-Chat, two small-sized versions of Tongyi Qianwen, will lower the threshold for the application of LLMs and let SMEs and developers start using AI quicker and faster, Alibaba Cloud said.

Lu Yanxia, research director at market consultancy IDC China, said the rollout of open-source AI models is expected to make AI services accessible to more enterprises, help them build customized industry-specific LLMs and promote the commercial use of LLMs, with an increasing number of Chinese tech companies likely to launch open-source models.

Other Chinese tech giants such as Baidu Inc, Tencent Holdings Ltd and iFlytek Co Ltd have been aggressively developing their own AI models in recent months.

In addition to unveiling open-source LLMs, Chinese tech companies should pool more resources into improving computing power, algorithms and quality of data, so as to gain a competitive edge in the AI chatbot race, said Pan Helin, co-director of the Digital Economy and Financial Innovation Research Center at Zhejiang University's International Business School.

LLMs refer to computer algorithms that are trained with huge amounts of data and are capable of generating content such as images, text, audio and video. They are the key technology underpinning AI chatbot ChatGPT.

As of May, China had developed at least 79 AI LLMs, or rivals of ChatGPT, according to a report released by the Institute of Scientific and Technical Information of China.