Alibaba's big language model topped the global ranking of AI platform Hugging Face

Three of the four Chinese LLMs in the top 10 spots in the rankings were from the Tongyi Qianwen series, also known as Qwen, developed by e-commerce and cloud computing giant Alibaba, according to AI and machine learning developer platform Hugging Face, which published an updated leaderboard with new figures on Wednesday. Alibaba owns the South China Morning Post newspaper.

"Qwen 72B [instruct] is king, and Chinese open models dominate overall," Hugging Face co-founder and CEO Clement Delang said Wednesday in a post on X, formerly Twitter.

Tongyi Qianwen's big language models, developed and managed by Alibaba Group Holding's cloud computing division, are used in industries ranging from consumer electronics to automobiles to online gaming.

This particular Qwen model was noted for its "success in math, long-range reasoning and knowledge," according to a post published Wednesday on the Hugging Face developer community website.

Meanwhile, Alibaba's Qwen-72B and Qwen1.5-110B models ranked third and tenth, respectively, as of Thursday. According to Hugging Face, the rankings are expected to change over time as more LLMs are evaluated.

The Yi-1.5-34B-Chat from Beijing-based startup 01.AI took seventh place. The company was founded in 2023 by renowned venture capitalist Li Kai-fu, who previously served as president of Google China.

Alibaba's high recognition of LLM is a testament to the company's rapid progress in artificial intelligence amid its strong commitment to open source development. Tongyi Qianwen has been open to third-party developers for several months now.

Open source provides public access to a program's source code, allowing third-party software developers to modify or redistribute its design, fix broken links, or extend its capabilities. Open source technologies have contributed immensely to the prosperity of China's technology industry over the past few decades.

An AI model from Facebook's parent company Meta Platforms ranked second behind Alibaba on Hugging Face's top 10 list. One LLM from Microsoft, a major sponsor of ChatGPT creator OpenAI, ranked sixth.

On Wednesday, the Hugging Face community said its methodology for ranking AI models has added additional metrics, such as complex tasks of around 1,000 words, while "AI performance is afloat."

Because the Hugging Face ranking focuses on programs developed in open source, it does not include advanced LLMs developed in a proprietary environment, such as OpenAI's GPT.