AI token cost
Work out what an AI feature costs to run each month in Australian dollars, at your own usage and exchange rate.
For Education organisations, RTOs, anyone budgeting an AI feature
How to use this calculator
Start with the worked example below. Replace the figures in Your numbers with your own inputs, then select Recalculate. Read the assumptions and limitations before using the estimate in your planning. Save report downloads the last calculated results and inputs.
Monthly cost (AUD)
$79.80
Input and output token cost converted at the rate you entered.
- Input cost (source currency)
- $12.50
- 10 million input tokens.
- Output cost (source currency)
- $40.00
- 4 million output tokens: usually the larger half of the bill.
- Cost per 1,000 requests (AUD)
- $15.96
- Useful for comparing models: per-request figures round away to nothing.
- Annual cost (AUD)
- $957.60
- Twelve months at this volume.
This is an estimate to support your professional judgement, not a determination. Our tools draft. You validate.
Saves the last calculated inputs, results and assumptions to your device. Recalculate after changing inputs.
Your numbers
How many times the feature calls the model in a month.
Roughly 0.75 words per token: a 1,500-word prompt is about 2,000 tokens.
Tokens the model generates back. Usually far fewer than the input.
From your provider's price list, in their currency.
Output is usually several times the input price.
Exchange rate. Use 1 if the price list is already in AUD.
How this is worked out
Input cost = (requests × input tokens ÷ 1,000,000) × input price. Output cost likewise. Monthly AUD = (input cost + output cost) × exchange rate.
What it assumes
- Token counts are averages across a month of real traffic.
- The exchange rate you entered holds for the period you are budgeting.
- Prices are quoted per million tokens, which is the common provider convention.
What it does NOT account for
- Does not include cached, batch or discounted pricing tiers.
- Does not include retries, failed calls or a fallback provider.
- Does not include hosting, storage, egress or staff time.
- Provider prices change without notice. Re-check them before committing to a budget.
What to do next
- Measure real token counts from a week of traffic rather than estimating them.
- Model a bad month at two or three times your expected volume.
- Check whether a cheaper model is adequate for your highest-volume request type.
- Re-check the provider's current price list before signing off a budget.
Sources
- Pricing model: Per-million-token input/output pricing, the common provider convention (reviewed 2026-08-26)