AI & Productivity

AI API Cost Calculator

Estimate monthly AI API spend from expected token volume, request count and per-model pricing.

Delivery: Instant download
Purchase: One-time payment
Format: Google Sheets
Excel
Updated: August 2026

$5.90Buy on Etsy

Sample preview — illustrative data

Cheapest at this volume

$15.54/mo

Gap to the dearest

$1,490/mo

ModelSaved by caching

Frontier model$1,505 / saved $441
Mid-tier model$301 / saved $88
Small fast model$82 / saved $24
No-caching model$304 / saved $0

What this spreadsheet does

Per-token pricing across models makes it hard to guess monthly API cost before a feature ships, and bills often arrive higher than expected.

What you can do with it

Estimate monthly cost before launching an AI feature
Compare projected cost across several models at the same volume
See what prompt caching is actually worth on your workload

What's included

Sheet tabs

Instructions
Cost Estimate
Model Comparison

Features

Separate input, output and cached-input pricing per model
Prompt caching modelled with a cache-hit rate you set
Monthly cost shown with and without caching, so the saving is a number
Models with no caching discount detected and counted
Cheapest and most expensive model at your volume, named automatically
A budget check: what your monthly budget actually buys

How it works

1Enter monthly requests and average input and output tokens per request
2Enter each model's prices per million tokens, including cached input
3Set your cache-hit rate and read the comparison

Who it's for

Builders and teams estimating or monitoring AI API spend.

Frequently asked questions

Do I need to update pricing manually?

Yes, and deliberately. Model pricing changes often, so rates are editable inputs rather than baked in — a spreadsheet that ships with prices hardcoded is wrong within a quarter.

Can I compare more than one model?

The Model Comparison tab prices every model you list at the same volume and names the cheapest and the most expensive, with the gap between them.

Does it handle prompt caching?

Yes, and it is usually the biggest lever. Input tokens are split by your cache-hit rate and priced separately. If a model has no caching discount, enter the same price for cached input and the sheet stops crediting you a saving.