emonster tools

AI product unit economics calculator

Published

Model monthly AI spend, cost per user, margin after entered costs, break-even usage, and lower or higher usage scenarios. Enter paying users, sessions and calls, then token usage, illustrative prices and operating costs in one currency. Check a subscription idea or stress-test high usage before pricing it.

Free to use. No email or account required.

Bookmark

Bookmark this tool

Save this link in your browser’s bookmarks.

Paste this link into the address of a new or existing bookmark.

On desktop, drag AI product unit economics calculator to your bookmarks bar.

Only the tool’s link is shared. Your inputs and results are not included.

Share

Share this tool

Copy this link to share this tool.

Only the tool’s link is shared. Your inputs and results are not included.

Your inputs

Edit the example inputs to match your task. Results update as you edit.

1–1,000,000,000
≥ 0
≥ 0
0–1,000,000,000
0–1,000,000,000
≥ 0 /1M
≥ 0 /1M
≥ 0 /1M
0–100 %
0–500 %
≥ 0
≥ 0
0–100 %
≥ 0
≥ 0

Calculations run in your browser. Inputs are not saved or sent to analytics.

How it works

Formula

  • Cost per attempt=Input tokens per call×(Input price per million tokens×(1−Input tokens served from cache100)+Cached input price per million tokens×Input tokens served from cache100)+Output tokens per call×Output price per million tokens1,000,000+Image, audio, and tool cost per attempt
  • AI cost per call=Cost per attempt×(1+Extra attempts per initial call100)
  • Monthly AI spend=Paying active users×Monthly sessions per user×AI calls per session×AI cost per call
  • Monthly surplus after entered costs=Monthly revenue−Payment processing cost−Other monthly costs−Monthly AI spend
Input and formula key

Your result

Complete the inputs to see your result.

Support

Sources

This calculator uses the method shown above. Results depend on the values and assumptions you enter.

How it works

Assumptions and limitations

All active users in this model pay the entered monthly price. Cache share applies only to input tokens. Retries are extra attempts per initial call, each with the same average usage and extra image/audio/tool cost. Costs scale linearly with sessions. The surplus includes entered costs only; it excludes taxes and unentered acquisition, payroll, or infrastructure costs. If revenue does not cover entered non-usage costs, no usage level breaks even. If those costs are covered and the cost per session is zero, usage adds no cost-based limit.

How to use this tool

Enter paying users, sessions and calls, then token usage, illustrative prices and operating costs in one currency. Compare the surplus and break-even when changing usage or cache share.

When this is useful

Check a subscription idea or stress-test high usage before pricing it. Prices are your inputs, not live provider quotations; free users and unentered acquisition or infrastructure costs are excluded.

Worked examples

  • Estimate a month of AI calls

    • With 100 users, 10 sessions each and 2 calls per session, there are 2,000 initial calls per month. Assume 1,000 input tokens at USD 1/million and 500 output tokens at USD 4/million, with no caching, retries or extra call costs. These prices are illustrative.
    • AI cost per call = (1,000 × 1 + 500 × 4) ÷ 1,000,000 = 0.003 USD; Monthly AI spend = 100 × 10 × 2 × 0.003 = 6 USD.

Frequently asked questions

What should I check before using the result from AI product unit economics calculator?

All active users in this model pay the entered monthly price. Cache share applies only to input tokens. Retries are extra attempts per initial call, each with the same average usage and extra image/audio/tool cost. Costs scale linearly with sessions. The surplus includes entered costs only; it excludes taxes and unentered acquisition, payroll, or infrastructure costs. If revenue does not cover entered non-usage costs, no usage level breaks even. If those costs are covered and the cost per session is zero, usage adds no cost-based limit.

Processing and privacy

The calculation or transformation runs in your browser. The tool does not send its input values or generated output to an application server. Loading a page still makes infrastructure requests; optional analytics and Google consent messages have separate controls.

Privacy Policy. Privacy & preferences in the footer controls optional analytics.

Help and feedback

Consult the linked sources for definitions and reference data. Contact Support if you have a question or notice a problem.

Read about calculation methods · Report a problem to emonster

Explore this subject: AI product economics

Your privacy & preferences

Optional analytics helps us understand how emonster.tools is used. The site works normally without it. Your analytics choice is stored on this device so you can change it later; choosing not to allow analytics prevents optional measurement and removes Google Analytics cookies this site can access. Read Privacy Policy