Token's Chinese Name Officially Determined
The Chinese name for "Token" appears to be finalized! Today, the State Council Information Office forwarded a People's Daily report citing an article from the National Data Administration, stating that the daily token call volume has exceeded 140 trillion. While 140 trillion data points is impressive, the Token's Chinese naming behind this news is more interesting, seemingly resolving recent industry disputes over the Chinese name for Token.
This also means the official Chinese name for Token has been set as "词元" (cíyuán).
The data released by the National Data Administration, published by the People's Daily, and forwarded by the State Council Information Office – a dual release by official government agencies and official media – solidifies the Chinese name for Token.
The naming dispute was sparked by the "模元" (móyuán) translation proposed by scholar Yang Bin in an article published by the Tencent Research Institute. This ignited a wave of discussion, with AI leaders on social platforms and industry groups offering their preferred answers. Soon, "智元" (zhìyuán) emerged and quickly gained widespread recognition.
You Yang, Young Professor at the National University of Singapore and Founder of LuChen Technology:
“Calling Token ‘智元’ is interesting. I wonder if the founder of ‘新智元’ (New Intelligence Yuan) has traveled through time.”
Hu Yulin, Independent Scholar and Former Associate Professor of History of Science at Tsinghua University:
“‘智元’ is actually better. The unit of a computer is a byte, while Token is the computational unit of human knowledge. Token is not a concept of general computation, but a concept of ‘general intelligence.’ ‘智’ (intelligence) is more to the point than ‘通’ (general).”
Wang Xiaochuan, Founder of Baichuan Intelligent:
“Calling it ‘智元’ is pretty good.”
Chen Xu, Senior Open Source Contributor:
“I’ve seen various ‘x元’ proposed these past few days, but only ‘智元’ resonates with me and meets the requirements of accuracy, clarity, and elegance.”
This large-scale discussion also attracted some playful participants.
But ultimately, the official decision is made: "词元"!
According to People's Daily, "词元" is the smallest unit of AI understanding of human language. The units split by token are sometimes characters and sometimes words, and words better cover this range.