How You Pay for AI: Subscriptions vs the API

You've probably heard two things about what AI costs, and they flatly contradict each other. One: it's a flat monthly fee, like a streaming subscription. Two: you "pay per word" — every message burns a little fuel, and it all adds up. So which is true?

Both. They're two different products wearing the same name. And once you can tell them apart, the pricing stops being confusing.

Back when we mapped the tools, I mentioned a pay-as-you-go option called the API and said we'd come back to it. Here we are. The easiest way to hold the whole thing in your head is one image: a transit pass versus a taxi meter.

The transit pass: a subscription

Most people pay for AI the way they pay for a monthly transit pass: one flat price, hop on as often as you reasonably want. That's the consumer subscription — a fixed monthly fee for the chat app, the thing you open and type into. Most of the big assistants also have a free tier, so you can ride a little before you buy the pass at all.

Here's where people get tangled, and it's the whole point of this lesson. On a transit pass, you're not paying per ride — you already paid; you just can't ride endlessly at rush hour. A subscription works the same way. You're not metering out fuel with every message. You pay the flat fee, and once in a while you bump a usage limit (a cap on how much you can send in a stretch) and wait a bit. That's it.

One quick variant before we cross over: teams and companies pay on the same flat-fee idea, just per person — one paid login each, called a seat. If you're here on your own, that's all it is — the same pass, bought in bulk.

The taxi meter: the API

Now the other product. The API is the genuinely metered one — you pay only for what you use, measured by how much text goes in and comes out (those tokens you've heard mentioned). Picture a taxi meter: it ticks up with every trip. Cheap for a short hop; pricey if you ride around all day.

Who's it for? This is the builder's door. It's the plumbing developers wire into their own apps — and it's what those multi-model aggregators from the last lesson are quietly running on underneath. You can get an API key and plug it into other tools, but if that sentence lands as gibberish, good news: you're not the audience for it, and you don't need to be.

"But isn't the API cheaper?"

You may hear that the API is the cheaper, savvier way to pay. Sometimes — but not as a rule. Use AI rarely, and paying only for the few times you do can come out cheaper than a monthly pass. Use it heavily all day, and that meter keeps ticking and can cost far more than the flat fee. Predictable flat price for steady use; pay-as-you-go for light use or for building. Neither one is the "smart" choice — they just fit different trips.

There is one real "it costs more" case worth knowing. Remember the aggregators that bundle many models into one app? They run on top of these APIs, and some add a markup for the convenience — so going through them can cost more than going to a provider directly. Convenience has a price; just know it's there.

So what should you actually do?

That settles the money side: what AI is, how to talk to it, which tool to open, and now what it costs. There's one more thing worth seeing before you go, and it's the fun one. This whole time we've been talking about words — and words turn out to be just the start of what these things can make. That's next.