A five-turn support chat costs $0.0168 per request on Claude Sonnet 5.5 with no optimisation, and 79 percent of that bill is input tokens (our model, 2026).
Prompt caching cuts the request to $0.0102 on multi-turn chats but adds a 13.5 percent penalty on one-turn calls, and routing 70 percent of requests to Haiku 5.5 reaches $0.0035.
On a $29 seat, break-even moves from 1,726 requests per month on naive Sonnet 5.5 to 8,278 with caching and routing, while naive Fable 5.1 breaks even at 345.