Skip to content

AI Cost Engineering

Tokens × price × volume, and the levers that cut it 10×.

token pricingprompt cachingcache ratiomodel routingmodel cascades

Lessons

  1. 35.1The LLM cost model: from tokens to a monthly billHow a few simple numbers turn into your monthly AI bill, so you can predict it instead of being shocked by it. 30 min AI cost simulator
  2. 35.2Routing for cost: small models for most, large models where it mattersMost questions are easy. Send them to a cheap model, save the expensive one for the hard ones, and check you didn't break anything. 30 min AI cost simulator Trade-off matrix

On the job: try it on your own project

Ask Claude Code to estimate your monthly LLM spend per feature from your logs or code: calls per day, tokens per call, model price and cache hit rate. Find the most expensive feature. Write a cost-reduction proposal for it: a cheaper model for easy cases, a cascade, prompt caching or shorter prompts, with the expected saving and how you will check quality does not drop.

Can you do this now?

0/4. Anything you can't tick yet: reopen that lesson's practice ladder.