AI Agent Cost Calculator
Estimate multi-step agent spend. Effective steps = steps × (1 + retries) × max(1, tool calls). Tokens per task = effective steps × tokens per step (default 70% input / 30% output). Cost uses model $/1M; multiply by tasks per month for a monthly total. Estimates only; official model pricing prevails.
Math runs in your browser. Nothing is uploaded.
Tokens per task
—
USD per task
—
Monthly USD
—
Estimates only; official vendor pricing prevails (2026-10-30). Effective steps = steps × (1 + retries) × max(1, tools); tokens/task = effective × tokens/step (70% in / 30% out).
RAG Cost Calculator · LLM API Cost Calculator · Grok Token Cost Calculator · OpenAI Token Cost Calculator
More in this group
- Transactional Email API Cost Comparison Amazon SES vs Resend vs Mailgun vs Postmark for one volume.
- Mailgun Cost Calculator Mailgun emails → plan, overage, validations and monthly $.
- Postmark Cost Calculator Postmark emails → cheapest plan, overage and monthly $.
- Amazon SES Cost Calculator Amazon SES emails × size → monthly cost and $/1k.
- Resend Cost Calculator Resend emails → cheapest plan, overage and monthly cost.
- LLM Observability Cost Comparison Langfuse vs Braintrust vs LangSmith vs Helicone for one workload.
- Langfuse Cost Calculator Langfuse units (traces + observations + scores) → monthly cost.
- Braintrust Cost Calculator Braintrust GB processed + scores + retention → monthly cost.
How it works
- How is AI agent cost calculated?
- Effective steps = steps × (1 + retries) × max(1, tool calls per step). Tokens/task = effective steps × tokens per step. Default split is 70% input / 30% output, priced at model $/1M. Monthly = per-task × tasks/month.
- Why multiply by tool calls?
- Each tool round often re-reads context. Treating tool calls as a multiplier is a planning heuristic—set it to 1 if your traces show otherwise.
- What do retries mean here?
- Retries add failed or repeated loops: factor (1 + retries). Zero retries means a single pass.
- Can I change the input/output split?
- The UI uses a 70/30 planning default. For exact bills, export token counts from your traces into the LLM or vendor token calculators.
- Are agent tool fees included?
- Only token spend at the model rates you enter. Separate tool SKUs (search, browsers, sandboxes) need their own line items.