Li Jiaqi's livestream sells Tokens: 100,000 Qwen Office monthly cards sold in half an hour

Under the AI wave, during this year's "Double 11," Token (word units) were listed as a commodity for the first time and became a sales entry point that major platforms are racing to stake out. On the evening of November 2, during Tmall's "Double 11" shopping festival, Li Jiaqi's livestream sold AI computing power products for the first time. Chen Yusen, CEO of Qianwen Office, appeared in the livestream to showcase Qianwen Office's application scenarios together with Li Jiaqi. It is understood that in about half an hour of sales, Li Jiaqi's livestream had sold 100,000 Qianwen Office personal membership monthly cards, totaling about 2 trillion Tokens. However, these 100,000 AI office personal membership monthly cards were priced at only 1 cent each. At the scene, Chen Yusen and Li Jiaqi also demonstrated multiple application scenarios of Qianwen Office in daily life and work—parents can use Qianwen Office to organize their children's homework and test papers and create targeted training mini-games; livestream operator Wangwang handed livestream replays and bullet comments to Qianwen Office to generate a business analysis report with one click... Li Jiaqi also told everyone that every AI generation and every long-text comprehension consumes Tokens, and this package does not buy a physical object but computing power quota that can help workplace professionals save time. Some fans said they had not used AI office tools before, and the host and guest mentioned many scenarios in which they could be used, which was quite tempting, so they bought it to try. Taobao begins selling AI computing power products. Many people surely remember that during "Double 11" in 2018, Li Jiaqi sold 15,000 lipsticks in 15 minutes, becoming a "famous scene." Lipstick is physical emotional consumption, while Token is instrumental digital consumption; the conversion logic of the two is completely different. According to the "Large Model Word Unit Service Performance Monitoring Report (2026)" recently released by the China Academy of Information and Communications Technology, in June this year, the average daily domestic consumption of word units was about 500 trillion, nearly four times that of December last year; the normalized average price rose from 6.8 yuan in the second half of last year to 13.46 yuan this year, an increase of nearly 100%. With both volume and price rising, Token appears to have become hard currency. In the same reporting period, however, the Silicon Data Token expenditure index, which measures real payment prices in the global market, fell below $1 per million Tokens, halving from its May high. One doubled, the other halved, but both are real. The difference is that what rose was the bill, not the unit price. In the past, when you asked AI a question, it consumed a few thousand Tokens; now, when you hand a task to an agent, it breaks down the steps itself, checks materials, and repeatedly verifies, consuming tens of thousands to hundreds of thousands of Tokens for one task. The share of usage of high-priced models, long context, and complex reasoning has risen sharply, pushing up the structural average price. The same type of task is becoming more expensive because you are making it do more things, not because computing power itself has become scarcer. You should know that the direction of technological progress is to make the same thing consume less and cost less. Recently, some domestic Token service providers have connected the "model—inference—chip" loop, doubling effective output under the same hardware investment and greatly reducing overall costs; some AI infrastructure companies have also cut Token costs to one-tenth of the original over the past year... Domestic inference efficiency is climbing, and the "gold content" of unit computing power is increasing. In such a market, the action of "stockpiling computing power" seems to make little sense—this is not a scarce resource, but merely a unit of measurement that is being continuously diluted by technology. Moreover, Tokens cannot be refunded or transferred—the measurement standards of various models are not unified at all, and they easily expire. IC image. So what is the livestream selling? It is selling extremely low-cost user acquisition and a psychological "getting on board." One cent requires no weighing and no price comparison; it precisely bypasses rational judgment. The number "2 trillion" detached from the context of "1 cent" in dissemination and became a stimulant: the AI era has really arrived, Tokens can even be snatched up in livestreams, and you still have not gotten on board? The livestream said that "AI cannot replace Li Jiaqi," which sounds like giving the host face, but in fact completed a clever dimensional reduction: AI is defined as a trivial tool that helps you get off work on time, accompany cats and dogs, and take care of family members, rather than something you need to seriously understand and use prudently. Fear was soothed, and the imaginative space for AI was narrowed accordingly. One cent does not buy efficiency. Efficiency is "finishing one thing in less time," while a person who has never opened Qianwen Office will most likely not know what to do with it after getting the monthly card. What it buys is the illusion of "already on board," a brief reassurance that "I have not been left behind," and the same sense of security as "buying a book means having read it, and getting a gym card means having gotten thin"—it belongs to the same psychological account. The ticket to the AI era has never been in the Double 11 shopping cart. What can truly get you "on board" is not how many Tokens you have stockpiled, but whether you have a problem for which you must use it, a workflow that works smoothly for you, and the judgment to know when it is reliable and when it is not.