Wise Agents
ENDE
Book a call
  1. Home
  2. Free tools
  3. Agent cost calculator
Free tool · Running costs

AI agent cost calculator: what one agent costs to run each month.

From your volumes and your model provider’s list prices: the model cost per month and per conversation, a starting daily cost cap, and the time your people spend reviewing. It runs in your browser and sends nothing.

Every figure here is an estimate. The default values are examples, not quotes or measurements. Replace them with your own volumes and your provider’s current prices.

Your agent

Volume
Example value. One customer chat, one claim or one document checked is one run.
Example value. Each turn of a chat, and each step of a longer task, is usually one call.
Size of each call
Example value: the instructions, the conversation so far and the passages it reads. As a rough rule, 1,000 tokens are about 750 English words.
Example value: the answer or draft it writes.
Your provider’s list prices
Example price, not a quote. Use your provider’s current list price for the model you test.
Example price, not a quote.
People in the loop
Example value. Every action that matters waits for a person; set the share that needs a decision.
Example value. Your own loaded cost per hour.
How it calculates

Four formulas, nothing hidden.

  1. Model cost per run = calls per run × (input tokens × input price + output tokens × output price) ÷ 1,000,000.
  2. Model cost a month = runs a month × model cost per run.
  3. A starting daily cap = 1.5 × (model cost a month ÷ 30). On the platform, a reached cap refuses new work until the next day; work already admitted is not cut off.
  4. Review time = runs × share reviewed × minutes per review ÷ 60, and its cost at your hourly rate.
What it leaves out

Costs a real proposal adds

  • The buildA pilot agent takes 6–10 weeks, a production agent 3–5 months. We don’t publish prices: book a call and the proposal gives a fixed price for your scope.
  • Tests and re-testsEach test round runs your cases through the model too. It is small next to live volume, but not zero.
  • Retries and long documentsA long contract or a retried step can cost many times an average call. The per-run limit on the platform caps that.
Pricing and what moves it AI agent monitoring and operations
Questions

About the calculator

Where do the default prices come from?

They are example values to show how the calculation works, not any provider’s price. Model prices change often and differ by model, so enter the current list price of the model you plan to test.

Why set a daily cap at all?

Because a loop, a long document or a spike in traffic can multiply costs in hours. On the platform, daily and per-run caps refuse new work when reached, and a person can raise them. The calculator suggests a starting point.

Does the calculator send my figures anywhere?

No. It runs entirely in your browser, with no form submission, no cookie and no analytics.

Bring one process. Leave with a risk class, a pass bar and a plan.

A 45-minute scoping call with the people who would build it. You get a first view of the AI Act risk class, the controls it needs, and whether an agent is the right tool at all.

Book a scoping call Try the demo

Or read the trust page first