picking the model: intelligence for the price

the lever you keep forgetting to pull

Module 0 covered what a session costs; this whole module has been about spending context and tokens well. One lever moves the bill more than all the others combined: which model runs the session. The spread is not subtle: per the official pricing table, at the time of writing, Haiku costs $1/$5 per million input/output tokens, Sonnet $3/$15, Opus $5/$25, Fable $10/$50. A 10x range for the same conversation, and prices shift as models ship. Always read them from the live table, not from a course.