AI Model Cost & Capacity Estimator

Runs in your browser

Use this when: estimating token costs and peak load from your own assumptions—not checking live provider pricing or guaranteed capacity.

WHAT YOU'LL ACHIEVE

Compare volume scenarios, budget exposure and a throughput headroom proxy without live prices or provider-capacity claims.

THE TOOL: ASSUMPTION-BASED ESTIMATOR

Enter your own inputs.

The estimate is only as good as your assumptions. It's a guide, not a quote.

Prices are illustrative assumptions only. No model/provider is selected and no live pricing or capacity is queried.

Currency codeExample: GBP. This is a label, not a currency conversion.

Example prices are fictional planning inputs, not a live quote. Loading them replaces the current fields.

Illustrative input priceCurrency per 1 million input tokens; enter a quote or your own assumption.Illustrative output priceCurrency per 1 million output tokens; not fetched from a provider.Input tokens per requestAverage prompt/context token estimate.Output tokens per requestAverage generated-token estimate.Calls per user per monthYour expected request volume before retries.Number of usersExpected users in the modelled month.Peak requests per secondYour peak-load assumption; this does not reveal provider capacity.Target utilisation (%)Your chosen headroom assumption, from above 0 to 100%.Retry rate (%)Expected additional requests as a percentage of base volume.Growth assumption (%)Expected volume growth for this scenario.Monthly budget limitOptional user-entered budget in the selected currency.

Enter valid non-negative assumptions:

Illustrative input price needs a number.
Illustrative output price needs a number.
Input tokens per request needs a number.
Output tokens per request needs a number.
Calls per user per month needs a number.
Number of users needs a number.
Peak requests per second needs a number.
Target utilisation (%) needs a number.
Retry rate (%) needs a number.
Growth assumption (%) needs a number.
Target utilisation must be greater than 0% and no more than 100%.

SCENARIOS AND CAPACITY

Read the estimate with its assumptions.

Results are withheld until every required numeric input and currency label is valid.

Formula: per request = input tokens ÷ 1,000,000 × input price + output tokens ÷ 1,000,000 × output price. Modeled requests = users × calls/user × (1 + retries) × (1 + growth). Monthly estimate = per-request cost × modelled requests. All figures exclude taxes, fees, discounts and other provider billing rules.

EXPORT OR RESET

Keep the assumptions with the result.

Copy or download includes units, formulas, scenarios and uncertainty notes so estimates are not mistaken for quotes.

AI MODEL COST & CAPACITY ESTIMATE

Currency: GBP
Input price assumption:  per 1,000,000 input tokens (illustrative/user-entered; not live)
Output price assumption:  per 1,000,000 output tokens (illustrative/user-entered; not live)
Input / output tokens per request:  / 
Calls per user/month: 
Users: 
Peak requests/second: 
Target utilisation: %
Retry assumption: %
Growth assumption: %

RESULTS WITHHELD: correct all visible validation errors before estimating.

Uncertainty: prices, token counts, retries, growth and load are assumptions you supplied (or example inputs you explicitly loaded). Actual usage, provider billing, throughput, latency, limits and capacity can differ. No provider price or capacity is fetched or asserted. Not a quote or spending guarantee. Inputs remain in this browser session only.