Author: Kuli, Shenchao TechFlow
Recently, you may have noticed something: people have started discussing what Token should be called.
Professor Yang Bin from Tsinghua University published an article titled "It's Urgent to Determine the Chinese Translation for Token"; a related translation question on Zhihu garnered 250,000 views, with comment sections flooded with suggestions.
Over the past two to three years, the domestic AI circle directly used the term "Token" without any issue. Why the sudden need for a Chinese name?
The immediate reason might be that, after this year's Spring Festival, ordinary people learned for the first time that Tokens cost money.
OpenClaw transformed AI from chatting to working, with a single task burning through hundreds of thousands of Tokens, sending bills skyrocketing; various cloud providers have also announced price increases, with the billing unit being Token.
At the same time, Token started appearing in places it hadn't before.
At the GTC conference, NVIDIA CEO Jensen Huang mentioned that in Silicon Valley, people are now asking in job interviews, "How many Tokens does this job offer?" He suggested incorporating Tokens into engineers' compensation;
OpenAI founder Sam Altman took it even further, suggesting that Tokens would replace universal basic income, with everyone receiving computing power instead of money.
Data from the National Data Bureau shows that China's daily Token consumption surged from 100 billion in early 2024 to over 40 trillion by September 2025, reaching 180 trillion this February. At the beginning of the year, People's Daily published an article titled "A Casual Talk on Ciyuan (词元)" to explain the term to readers.
Once a technical term enters cloud service bills, recruitment compensation packages, and official statistical metrics, it can no longer remain in English.
The question is, what to call it?
If this were merely a translation issue, there would already be an answer. In 2021, the domestic academic community settled on a name for Token: 词元 (Ciyuan).
But no one cared back then because Token was still an internal term within technical circles.
Now, it's different.
The word "Token" itself is a versatile container; previously, people in the crypto sphere called it 代币 (Daibi, meaning token/coin), those in security called it 令牌 (Lingpai, meaning token/pass), and those in AI called it 词元 (Ciyuan, meaning lexical unit). The same English word, depending on which direction the Chinese translation leans, determines whose territory it belongs to.
Thus, a battle over naming Token began.
Business Needs Discourse Power
How a word is translated is usually a matter for linguists. But this time, almost no linguists are involved in the naming.
The currently most prominent name is "智元" (Zhiyuan).
It is being pushed most vigorously by an AI media outlet called "新智元" (Xin Zhiyuan). If the Chinese name for Token is set as "智元", this company's brand name would coincide with a fundamental industry term, effectively getting free advertising in every article discussing Token.
Their own promotional article ends quite frankly: "We suggest translating Token as the industry's new consensus: 智元 (Zhiyuan), leaving the '新' (Xin, meaning new) for us."
According to the same article, Baichuan Intelligent founder Wang Xiaochuan commented: "Calling it Zhiyuan is quite good."
As a maker of large models, it's certainly good for him if Token is called Zhiyuan. Each operation of the model would then produce not just a billing unit, but a "basic unit of intelligence."
Selling Token is selling traffic; selling Zhiyuan is selling intelligence—the valuation story is entirely different.
Professor Yang Bin from Tsinghua University proposed "模元" (Moyuan), with "模" (Mo) corresponding to model. Whoever owns the large model holds the production rights to "模元". Leaning the name towards models shifts pricing power to the model companies.
Some advocate for "符元" (Fuyuan), returning to the most fundamental definition in computer science—Token is simply a unit of symbolic processing, unrelated to intelligence or models.
It's the cleanest technically, but the proposer is an independent technical writer, without corporate backing or capital push, and thus has almost no voice in this discussion.
Whichever direction the name leans, the industry narrative moves that way, and money flows accordingly.
A distant example: the day Facebook renamed itself Meta, "metaverse" transformed from a sci-fi concept into a valuation story for a company. A recent example: China consumes 180 trillion Tokens daily, ranking first globally, but what to call it, how to define it, and who defines it remain undecided...
The world's largest consumer of Tokens hasn't even decided what to call what it consumes.
However, this term actually already had a Chinese name.
In 2021, Professor Qiu Xipeng from the School of Computer Science at Fudan University translated Token as "词元" (Ciyuan). The academic community accepted it and wrote it into textbooks. Nobody discussed it then because Token wasn't valuable back then.
Now Token is valuable.
It is the billing unit for cloud services, the revenue source for large model companies, and a core metric for national statistics on the AI industry's scale. So the media arrived, the big shots arrived, the professors arrived, each bringing their preferred name and the rationale behind it.
Translation was never the problem. The problem is when this term started becoming valuable.
Jensen Huang did not participate in the Chinese naming discussion at GTC. He did something simpler: held up a championship belt printed with "Token King" and declared that data centers are Token factories.
Whoever produces Tokens, defines Tokens. What the name is, he doesn't care.
Token, Land Grabbing, and Coin Minting
Therefore, the part truly deserving serious thought in this matter is not which translation is better.
After the term "calorie" was established, the entire food industry's pricing, labeling, and regulatory systems were built around it. After the definition of "流量" (Liuliang, data traffic) was established in China's telecommunications industry, operators billed, competed, and designed packages based on it—the entire business model revolved around these two words for over a decade.
Token is now on the same path.
It is already the billing unit for cloud services, the revenue metric for large model companies, and a core indicator for measuring the AI industry's scale at the national level. The VC circle is even starting to talk about whether investment disbursements can be made directly in Tokens.
Once a word becomes a measure of money, naming it is no longer translation; it's minting currency.
Call it "智元" (Zhiyuan), and the minting right belongs to the AI narrative; whoever tells the story of intelligence benefits. Call it "模元" (Moyuan), and the minting right goes to the model companies; whoever has the large model prints money. Call it "符元" (Fuyuan), and the minting right returns to the technology itself, but technology itself doesn't speak for itself.
The academic community's 2021 term "词元" (Ciyuan) was ignored not because the translation was poor, but because this "coin" wasn't valuable back then.
Now it's valuable, and everyone wants to carve their name on it.
