Why AI Pricing Feels Weird (Per-Use vs Flat)

Lesson 2 · taught by Pip · free preview

Most things you buy have a price you can memorize. A coffee is $4. A haircut is $30. But a lot of AI is priced like your electric bill — it charges for how much you use it, and the number moves every month. That's why an AI bill can feel slippery. This lesson makes it predictable.

The plain idea: AI usually comes in one of three pricing shapes, and once you can name the shape, you can predict the bill. No shape is a trap by itself. The trap is not knowing which one you're standing on.

The three shapes

Flat subscription. You pay the same amount every month no matter how much you use it — like a gym membership. Easy to budget. The risk runs the other way: you might pay for capacity you never touch.

Per-use (usage-based). You pay for what you consume — each conversation, each document, each message. Like a taxi meter or your water bill. Cheap when you're quiet, but it climbs with success. Busy month? Bigger bill. This is the shape that surprises people, because the cost grows exactly when business is good.

Tiered. A blend: a flat base price that includes a bucket of usage, then per-use charges once you exceed the bucket. Like a phone plan with "5,000 minutes, then overage." Comfortable until you blow past the bucket, then it stings.

Why does AI lean toward per-use so often? Because the company providing the underlying intelligence charges its sellers by the amount of work done — every answer costs them a sliver of computing power. That cost gets passed down to you. It's the same reason a print shop charges per page: each page genuinely costs them something.

An everyday picture: gym vs taxi

A gym membership is flat — $40 whether you go thirty times or zero. A taxi is per-use — the meter runs the whole ride. Neither is "better." If you go to the gym every day, the membership is a steal. If you take one taxi a year, paying per ride is obviously smarter.

The whole game is matching the shape to how you'll actually use it. Heavy, steady use loves a flat rate. Light or spiky use loves per-use. Pick the wrong shape for your pattern and you overpay — not because the price was unfair, but because it was the wrong shape for you.

Your turn

For any AI tool you're looking at, ask the seller one question: "Is this flat, per-use, or tiered — and what makes the bill go up?" That last half is the important half. If they can't tell you plainly what makes your bill grow, you don't yet know what you're buying.

🔦 You can now read the shape of a price. Next, we'll turn that into something you can plan around: a simple, honest AI budget.