THE COST OF USEFUL AI / LABS 02 AND 03

What does the next call cost?

Fictional teaching rates, not provider quotes. These calculations run in your browser.

Keep each category separate

Each token belongs to one input category. The write rate replaces ordinary input pricing for those tokens. Human handling, fixed operations and implementation are outside this receipt.

Ordinary input

tokens

Cache write

tokens

Cache read

tokens

Output

tokens

Ordinary input rate

USD / million tokens

Cache write rate

USD / million tokens

Cache read rate

USD / million tokens

Output rate

USD / million tokens

Retrieval

USD / request

Tools

USD / request

USD / REQUEST

$0.0399

Itemized request cost in US dollars
Line	Cost
input	$0.0040
write	$0.0075
read	$0.0010
output	$0.0064
retrieval	$0.0060
tools	$0.0150

Model subtotal: $0.0189. Retrieval and tools are added once.

Try Chapter 5

Reset, then change output tokens from 800 to 400. The total becomes $0.0367, saving $0.0032 per request. Check the effect on acceptance and human handling before calling it a workflow saving.