Sunk Cost sunkcost.ai

If you buy a machine to run local models, how long until it pays for itself, and what can it actually do?

Data checked

Your machine

I’m looking at a , the with of memory.

About this machine

The break-even

Post on X

How you’d use it

Assumptions you can change

Where nothing has been measured, local speed is estimated as memory bandwidth ÷ bytes read per token × , and labelled as such. API speed only affects the time comparison.

How this is calculated

The small print that isn't small