Lesson 1 of 7
Two billing worlds
6 min read
The chatbot says 'free', or maybe you pay a flat $20 a month. So who actually pays for all that computation — and how?
Flat fee vs metered
There are two ways to pay for AI. A consumer subscription is a flat monthly fee with a usage cap — simple and predictable, but throttled once you hit the limit. The [API](glossary://api) is metered: you pay per token for exactly what you use, with no cap — that's how apps and developers buy AI, and how the 'free' tier is really funded behind the scenes.
Subscription = flat and capped. API = metered and uncapped.
Which one is cheaper?
It depends entirely on how much you use. For light, occasional use the metered API can cost pennies — less than a subscription. For heavy, constant use the API bill climbs without limit, while the subscription stays flat (until its cap throttles you). Neither is 'cheaper' in the abstract; the right question is always at my volume.
There's no cheapest option in general — only cheapest for your volume.
When you see an AI price, first ask which world it's in: a flat subscription you can't exceed, or a metered rate that scales with every token.
The two worlds
- —Subscription: flat monthly fee, usage capped.
- —API: metered per token, no cap.
- —Cheapest depends on your volume, not on the label.
What's the core difference between a subscription and API pricing?
Continue in the app
Take the whole What AI Costs course — tracked
Get your personalized path, progress and streaks in the app — this lesson and every next one, in order.