AI Agents
·
·
3 min read New
How I use several AI models to code without the bill blowing up
Almost 30 billion tokens in six months, and 94% of it was cache re-reads. How I split the work between one expensive model and several cheap ones.
delegation caching tokens workflow
