Short answer: an agent run costs its model tokens, plus paid tool and browser calls, plus the time a person spends reviewing it, plus the expected cost of redoing runs that fail. Put your own numbers in below. Nothing is sent anywhere: the math runs in your browser, and the link you copy holds your inputs.
Your numbers
The starting values are the illustrative example from our cost-per-run guide, not benchmarks or current list prices. Use your provider's current price sheet and numbers from your own logged runs.
Cost per run
| Part | Per run | Share |
|---|
If tokens, review time and rework each swing 50% either way, the monthly cost lands at . Fixed per-run tool and browser costs are held steady.
Biggest levers
The formula
cost per run = token cost + tool/API cost + browser/compute cost + (review minutes × hourly rate ÷ 60) + (rework rate × cost of a redo)
Token cost is input tokens × input price plus output tokens × output price, summed over every model call in the run. In most business workflows the review minutes outweigh the tokens, so cutting a minute of checking often saves more than switching models. The full method, with a worked example, is in How to estimate an AI agent's cost per run.
Need the full model?
The Cost & ROI Calculator ($29, Excel) adds run types, a browser surcharge, fixed monthly costs, low/base/high scenarios and a calibration tab for your logged runs. 14-day no-questions refund.
Related free tool: MCP config audit, which checks an MCP config for risky settings in your browser.
Product names on this page are trademarks of their respective owners and are used only to say what Agent Field Kit works with. Agent Field Kit is independent and is not affiliated with, sponsored or endorsed by them.