https://j.people.com.cn/n3/2026/0327/c95952-20440682.html

A Ciyuan (Token) is the smallest unit used by large language models LLMs to process information, and in the age of AI, it is quantifiable, priced, and tradable. According to Associate Professor Ma Zhiheng of the School of Computational Power and Microelectronics at Shenzhen University of Technology, before being input into the model, text is divided into tokens and converted into vectors. For example, in Chinese, typically one character corresponds to one or two tokens, and a certain amount of tokens are consumed each time a user asks a question or the AI ​​answers.

“Token” are also the evaluation criterion for the call volume of Large-Scale Language Models (LLMs) in rankings, and also the billing unit in the sales plans of LLM companies. “Token, or ‘Ciyuan,’ is not only a benchmark for value in the AI ​​era, but also a ‘settlement unit’ that connects technology supply and business demand, providing the possibility of quantification in the implementation of business models.”

In early 2024, China’s average daily token call volume was 100 billion, but by the end of 2025, it had jumped to 100 trillion. By March of this year, it had already surpassed 140 trillion, representing a growth of over 1000 times in two years, according to interviews with the National Data Administration.

The Chinese translation of Token has now formally be decided as “Ciyuan.”