A cloud software pricing structure where customers are billed based on the exact quantity of text fragments (tokens) processed or generated by an artificial intelligence API.

Brandished by AI infrastructure vendors to make variable usage pricing sound like a democratized utility, while disguising how rapidly recursive prompts and high-volume data runs drain client budgets.

• Unpredictable Metering: Masking variable monthly cloud costs behind microscopic fractions of a cent per word unit.
• Metering Mystique: Draping basic word-count billing in complex MLOps vocabulary.
• House-Cleaning Reality Check: Like paying a cleaning service by the exact number of carpet fibers swept rather than a flat room fee.

The Wolf of Wall Street (2010s) scene where Mark Hanna explains to Jordan Belfort how commission fees chip away at client capital on every micro-transaction while the house collects cash regardless of outcome.

Usage-based billing that charges customers per word fragment processed.

“Adopting a pay-per-token model keeps initial setup expenses minimal for lightweight apps.”

“We switched to usage billing, so our monthly cloud invoice goes up or down depending on how many words our users query.”