A free tool from Profit After Fees

Will your new AI feature still make money at scale?

Enter your model, your tokens per call, and what you charge per user. You'll see your real per-user AI cost, your gross margin at today's price, and the price you'd need to hit the margin you actually want.

  • Editable token-price presets.
  • Gross margin updates as you type.
  • Share a scenario with a link.

AI feature cost calculator

Your numbers

Model the feature before you price it

Everything stays in your browser. Nothing is saved on our side.

Model & tokens
Usage
Pricing target

Treat the model presets as a starting point, not gospel. Paste in your actual provider rates before you make pricing decisions.

Your verdict

AI cost against your price

Updates as you type. Negative numbers mean this feature loses money on every user it touches.

Verdict

Break-even monthly price

Below this, AI cost and fixed tooling eat the user's entire subscription.

Suggested price for target margin

What you'd need to charge to hit the margin you set above.

API cost per call
AI cost per user per month
Total monthly AI cost
Fixed cost per user
Gross margin at current price

How the AI cost math works

The calculator prices one request by multiplying prompt tokens by your input-token rate and output tokens by your output-token rate. Then it multiplies that per-call cost by calls per user per month.

Fixed tooling costs are spread across paying users, so the same feature can look painful at 10 users and healthy at 1,000. Break-even is the monthly price where AI cost and fixed tooling cancel out subscription revenue.

Suggested price is the monthly price needed to keep the gross margin you picked. Most products need to live above break-even because gross margin funds everything that is not tokens.

Common questions

Questions founders ask before they trust the number

Are these token prices the real prices my provider charges?

No. The model presets are example rates we filled in to save you typing. Provider pricing changes, varies by region, and depends on your contract. Open the input and output price fields and paste in what you actually pay.

What counts as a fixed monthly AI tooling cost?

Anything you pay every month regardless of how much the feature gets used: a managed vector database, embeddings pipeline, eval tooling, observability, or a reserved-capacity minimum.

What's the difference between break-even price and suggested price?

Break-even is where this feature stops losing you money. Suggested price is what you'd charge to keep the gross margin you set as your target.

My margin looks great at 100 users and terrible at 10. Why?

Fixed costs. A tool that costs the same at 10 users and 10,000 users gets cheaper per user as you grow. Variable token cost stays roughly constant per user.