If you tokenmaxx in the Meta offices and the leaderboard doesn’t register it, does it make a sound? Just days after The Information reported that Meta built an internal system to track which employees burn through the most tokens, the company has apparently taken down the leaderboard, leaving the entire office in the dark on who is pawning off the most work to AI assistants.
During its brief time on this earth, the leaderboard kept tabs on the top 250 employees across Meta who were using the most tokens, a unit of data that measures inputs and outputs from large language models. The higher a person’s token count, the higher they ranked on the scoreboard, where they were reportedly given badges like “Token Legend,” “Model Connoisseur,” “Cache Wizard,” and “Session Immortal.” The board was called Claudeonmics, named after Anthropic’s Claude model that has become the go-to model for vibe coding. That fact is a bit embarrassing for Meta, as its own models have fallen behind its competitors.
Per The Information, Meta employees certainly racked up some impressive figures. The outlet reported that the company’s staff used up about 60 trillion tokens in just 30 days, with the person at the top spot on the leaderboard having used 281 billion. For reference, the New York Times reported 210 billion tokens would produce the equivalent text needed to fill Wikipedia 33 times over.
We won’t know just how many tokens Meta is burning through anymore, though, since the company is suddenly feeling shy about the whole thing. According to The Information, the leaderboard has been replaced with a message that reads, “It was meant to be a fun way for people to look at tokens, but due to data from the dashboard being shared externally, we’ve made the decision to shutter Claudeonomics for now.”
Despite the lack of transparency on the burn rate, you can safely assume the company is still putting up numbers without the board. “Tokenmaxxing” is the new hotness in Silicon Valley, where the sheer volume of AI use has suddenly become a metric that is equated with productivity. Nvidia CEO Jensen Huang recently said that he would be “deeply alarmed” if an engineer didn’t use at least $250,000 worth of tokens per year. The Times reported that Shopify has started tracking employee AI usage, praising people who burn through tokens while chastising those who don’t. It also found an Anthropic engineer who reportedly spent $150,000 worth of tokens in just one month, apparently to the company’s pleasure.
Keep in mind, all of this is coming from an industry that is actively laying people off and cutting payroll in the name of “efficiency.” Suddenly, running up huge bills is perfectly acceptable despite a lack of any evidence that the output is any better.
Read the full article here
