ME News reports that on April 7 (UTC+8), an internal Meta leaderboard named "Claudeonomics," using the name of Anthropic's flagship product Claude, appeared on the company's intranet. Built by employees using company data, it aggregates token consumption from over 85,000 users and ranks the top 250. A copy of the leaderboard seen by The Information shows that total consumption over the past 30 days exceeded 60 trillion tokens; based on Claude Opus's publicly disclosed average price of approximately $15 per million tokens, this equates to roughly $900 million. However, Meta's actual model mix and pricing agreements remain unknown. The top individual user averaged 281 billion tokens consumed, with potential costs reaching millions of dollars.
The leaderboard features gamified incentives, progressing from bronze to emerald tiers, with top titles including "Token Legend" and "Session Immortal," as well as "Model Connoisseur" and "Cache Wizard." Some employees have had AI agents run research tasks for hours on end solely to inflate usage metrics. Neither Meta CEO Mark Zuckerberg nor CTO Andrew Bosworth made it into the top 250.
In Silicon Valley, a new trend called "tokenmaxxing" has recently emerged, with token consumption becoming a new metric for measuring engineer productivity. In February, Bosworth said at a tech conference that one top engineer’s token spending had reached parity with their salary, boosting productivity by up to 10 times: “It’s a no-brainer—keep spending, there’s no limit.” Last month, NVIDIA CEO Jensen Huang stated that if an engineer earning $500,000 per year spends less than $250,000 on tokens, he would “be deeply concerned.”
Meta engineers are currently using external models from Anthropic, OpenAI, and Google, along with internal tools MyClaw (Meta’s version of OpenClaw) and the recently acquired Manus. In an internal memo this year, Zuckerberg issued a “bold request” to the engineering team: rewrite Meta’s codebase so that AI agents can directly read and modify code.(Source: Meta)