Cost Engineering

1. What is considered the silent cost driver in token economics according to cost engineering?

Output tokens Context bloat Input tokens User engagement

2. Which caching strategy allows for reusing answers for nearly identical questions?

Provider prompt caching Semantic/response caching Embedding caches Contextual caching

3. What should be done to minimize costs related to input tokens?

Send all context with every call Use batch APIs for all calls Cache repeated prompts strategically Try the flagship model every time

4. What is the purpose of establishing budgets with quotas and alerts for features?

To limit overall spending on all features To alert users of high-quality model usage To surprise developers with unknown costs To track and control spending for each feature

Continue: Capacity, Latency, and SLOs → Review this lesson Retake quiz