Token Estimate

Free Token Estimate calculator with clear step-by-step results.

About Token Estimate

Prices an LLM workload with input and output tokens billed at different rates, which matters because output is typically three to five times the price of input. Prompt caching is applied as a discount on the input side only, since cached prefixes reduce input cost but never output cost.

Formulas

Effective input price

Price = list input price x (1 - cache hit rate x cache discount)

Daily cost

Cost = input tokens / 1,000,000 x effective input price + output tokens / 1,000,000 x output price

Worked example

With the default values below, Token Estimate returns Monthly API cost of $1,262.21. Change any input above and every figure here is recalculated in your browser.

Inputs used

Requests per dayminimum 0: 5,000
Input tokens per requestminimum 0: 1,200
Output tokens per requestminimum 0: 400
Input priceminimum 0 $/M tokens: 3 $/M tokens
Output priceminimum 0 $/M tokens: 15 $/M tokens
Prompt cache hit rateaccepts 0–100 %: 40 %
Discount on cached inputaccepts 0–100 %: 90 %

Results

Monthly API cost: $1,262.21
Daily API cost: $41.52
Tokens processed per day: 8.000 M
Cost per request: $0.008304

Step by step

1Values used
2Effective input price
3Daily cost
4Monthly API cost= $1,262.21
5Daily API cost= $41.52
6Tokens processed per day= 8.000 M
7Cost per request= $0.008304

Assumptions and limits

The result is derived only from the 7 inputs listed above. Anything not among them is not part of the calculation.
In particular, this calculator does not take inflation, tax, fees or charges or compounding frequency into account.
Figures are rounded for display to 6 decimal places, so totals built from several rounded parts can differ slightly from the unrounded arithmetic.

How these calculators are built and tested is described on the Methodology page. If a figure looks wrong, please report it.

Frequently Asked Questions

Why is output so much more expensive?

Output tokens are generated one at a time and cannot be batched the way a prompt can be processed in parallel, so each one costs far more compute to produce.

How much does caching actually save?

A great deal for repeated system prompts and long shared context - a 40% hit rate at a 90% discount cuts input cost by over a third. It does nothing for varied one-off prompts.