AI Agent Cost Calculator
Model the part naive token calculators miss: steps, retries and tools.
Estimate your workload
Change the assumptions on the left; read the modeled result on the right.
How to read this estimate
What this calculator is for
Agents rarely make one model call. Planning, tool selection, retries and validation turn one user task into many billed steps. This calculator makes that multiplier explicit.
How to read the result
Estimated monthly agent cost combines modeled LLM steps and paid tools. Cost per task divides that total by task volume. Billed LLM steps grows with the step count and retry rate; compare both to see which driver dominates.
The math
Monthly agent cost = (tasks × steps per task × (1 + retry rate) × model cost per step) + (tasks × tool calls per task × cost per tool call).
Pricing sources
Rates use the published TokenCOGS pricing registry and its September 16, 2026 fallback snapshot. Live values may differ; see methodology and source limitations before procurement.
Planning limitations
Actual invoices can differ because providers may apply long-context pricing, caching rules, batch rates, tool-call fees, regional processing, negotiated contracts, taxes, or future pricing changes. Treat the output as a planning estimate.
Use it to create scenarios and catch order-of-magnitude mistakes. Before committing spend, verify the selected model's live provider pricing and compare the estimate with actual usage telemetry.
Why can an agent cost more than a single model response?
A single task may require multiple calls, retries and third-party tools. All are billed even when the user sees one final answer.
Why do agent costs grow so quickly?
Each task may trigger multiple LLM calls and tool calls, and retries add more billed work. Long-lived context can increase input tokens further.