Credit-based pricing
Also known as: AI credits, prepaid credits, token credits, credit model
Instead of showing a bill based on tokens or API calls, many AI products sell credits: a unit of measure defined by the product itself. You buy a pack of 1,000 credits, and each generation, search, or task deducts some amount from your balance. Midjourney and ElevenLabs popularized this model at the consumer level; it is now widespread across prosumer and enterprise AI tools.
Credits work well because they give customers a sense of control. Buyers know upfront what they are committing to. Vendors benefit from cash collected before consumption, and can smooth out revenue even when usage is uneven. Credits also let products bundle different AI features at different costs under one system without exposing the underlying per-token math.
The downside is opacity. Enterprise finance teams often push back on credits because it is hard to forecast what a credit will buy in six months as models change. 'What exactly is a credit worth?' can become a deal-blocker. Products that use credits well tend to provide clear usage dashboards and refill nudges, so customers always know where they stand.