Free pricing brief

Turn AI usage cost into a price you can defend

Model cost is only the first line. Add failures, tools, fixed infrastructure, and your target gross margin to calculate a scenario price floor per successful task.

Imported workload

Customer Support Chatbot

using GPT-5.6 Terra

30,000 attempted tasks per month · model cost $0.03 per attempt

Model / successful task

$0.03

Fixed cost / success

$0.00370

Fully loaded / success

$0.03

Successful tasks / month

27,000

Scenario price floor

$0.11/ successful task

At this price and volume, the scenario produces $2,966 monthly revenue and $2,076 gross profit before tax and unentered costs.

Monthly scenario cost

$890

The link exposes these numeric assumptions to anyone who has it. It includes no name, email, or note.

Need plan prices and usage caps?

The planned Cost-to-Price Workbook extends this calculation into customer tiers, power-user stress tests, routing scenarios, and a client-ready decision report. The free calculation above remains free.

Review the workbook →
← Rebuild the workload comparison

Illustrative planning model, not financial, tax, or pricing advice.

What this price floor includes

The calculator starts with the selected workload's model cost per attempt. It divides that cost by the successful-task rate, adds the variable cost you entered, allocates monthly fixed costs across successful tasks, and then applies your target gross margin.

In formula form: price floor = fully loaded cost per successful task / (1 - target margin). The result is useful for scenario planning, but it is not a demand test and it does not replace your own cost ledger.

Why success rate changes the economics

A workflow that succeeds 80% of the time pays for 100 attempts but delivers only 80 outcomes. Retries, tool failures, malformed outputs, and human rejection can therefore make cost per delivered result much higher than cost per API call. Use production logs when you have them; use a conservative pilot estimate before launch.

Costs this calculator cannot know

Payment fees, taxes, refunds, discounts, customer support, sales time, founder labor, compliance, and unusually expensive users are not added automatically. Enter directly attributable variable and fixed costs, then stress-test the result before setting a public price.

FAQ

Should I use 70% gross margin for every AI product?

No. Margin targets depend on the product, support burden, market, and growth model. The default is a planning scenario, not an industry rule.

What if power users consume ten times more?

Run at least two scenarios: median usage and a high-usage stress case. A flat subscription may need a usage cap, overage price, or model-routing rule if a small group can erase the plan's margin.

Can I use the result in a client quote?

Use it as an internal starting point, then add delivery labor, risk, support, payment terms, and tax treatment. Keep the assumptions beside the quote so the price can be revised when usage changes.