How do I price an AI wrapper so it's profitable?

Start from your real cost to serve one user for a month — the API/LLM cost plus payment processing and any per-user infrastructure — then set a price that leaves the gross margin you need. For SaaS that is usually 70–85% gross margin, so if a user costs you $2/month in API calls you want to charge roughly $8–13/month, not $5. This calculator does the reverse-math: enter a target margin and it shows the minimum price.

Why do AI wrapper apps lose money?

Because the cost of goods is variable and invisible at signup. A flat $9/month plan looks fine until a power user makes 10× the average number of API calls and burns $15 of tokens — that single user is now unprofitable. Heavy users, no usage caps, and underpriced unlimited plans are the three classic ways an AI app's API bill quietly overtakes its revenue.

What is a good gross margin for an AI SaaS?

Traditional software runs 80–90% gross margin because the marginal cost is near zero. AI apps carry a real per-request cost, so margins are structurally lower — 60–80% is healthy, and below 50% leaves little room for support, sales and the inevitable power users. If your margin is thin, the fixes are usage caps, prompt caching, model routing to a cheaper model, or simply a higher price.

AI Wrapper Pricing Calculator

An AI app has a cost of goods — most founders forget it

Classic SaaS has almost zero marginal cost: once the software exists, the 1,000th user costs about the same as the 100th. AI apps broke that rule. Every prompt, embedding and generation is a metered charge from OpenAI, Anthropic, Google or whoever sits underneath you. That means your product has a real cost of goods sold per user, and your pricing has to clear it with room to spare. Charging $9/month for unlimited GPT-class output is the AI-era version of selling dollars for ninety cents — it scales straight into a bigger loss.

This calculator makes the hidden math visible. It subtracts the API cost, the payment processor's cut and any other per-user cost from your price to get the gross profit per user, then spreads your fixed costs across your user base to find the break-even point and your true net margin. Flip it around with the target-margin field and it tells you the minimum price to charge. If the number is uncomfortably high, that's the signal to cut the cost underneath rather than the price on top.

The three ways AI wrappers leak money

1. Power users on flat plans. Your "$4 average API cost" is an average — the heaviest 10% of users can cost 5–10× that. A flat unlimited plan lets them eat the margin of dozens of light users. Usage caps, credits or metered overage fix this. 2. Underpriced for the margin. If a user costs $4 and you charge $8, you're at a 50% gross margin before support and churn — too thin. 3. Ignoring the cost levers. Before raising prices, shrink the API bill: prompt caching, model routing to a cheaper model for easy requests, and the cheapest LLM API for your token mix can each cut cost of goods 30–70%, which drops straight to margin.

Size the cost before you set the price

The input that matters most is API cost per user, so measure it properly. Use the AI app cost estimator to turn calls-per-user and tokens-per-call into a monthly figure, the LLM token cost calculator for a single model, and the chatbot or RAG cost calculator if that's your shape. Then bring the number back here to price it. The order is always cost first, price second — guessing the price and hoping the cost fits is how the bill catches you at scale.

How to use it

1. Enter your monthly subscription price and the API cost to serve one user for a month.
2. Add payment fees (percent + fixed per charge) and any other per-user cost.
3. Enter your fixed monthly costs and current paying users.
4. Read the gross margin per user, net profit, break-even users — and set a target margin to get the minimum price you should charge.

Common mistakes

Pricing off the average user. Price so that an above-average user is still profitable, then cap the extremes. Forgetting payment fees. On a $9 plan, 2.9% + $0.30 is ~6% of revenue gone before API cost — material at low prices. Counting trials as users. Free and trial users carry API cost with zero revenue; model them separately. Treating margin as profit. Gross margin pays for support, sales, refunds and your time — leave room.

FAQ

How much should I charge for an AI feature?

Enough to keep a healthy gross margin over your real cost to serve a user. If API + fees cost you $3/user and you want a 75% margin, you need to charge about $12. Set the target-margin field and the tool gives the exact minimum.

What's a healthy margin for an AI app?

60–80% gross is healthy for AI SaaS — lower than pure software because of the per-request cost. Under 50% leaves little for support and power users; fix it with caps, caching, routing or price.

How do I stop power users from killing margin?

Add usage limits or metered overage, route their requests to a cheaper model, and cache repeated context. Model the savings with the caching and routing calculators before changing the price.

Should I charge per seat or per usage?

Per-seat is simpler but exposes you to heavy users; usage-based (or seat + included credits + overage) aligns price with cost of goods and protects margin. Many AI apps land on a hybrid.

Estimate only. API prices and payment fees change — verify current rates with your providers before pricing.

AI wrapper pricing calculator

Profit as you scale