2026-01-03
5 min
Prompt Caching: How to Get Cache Hits and Reduce Costs
Prompt caching can cut input token costs by 75%, but most apps get zero cache hits. Structure prompts correctly, measure cached_tokens, and stop re-paying for the same prefix.
2026-01-03
6 min
Cursor Model Selection: Cost vs Performance Breakdown
Cursor credits burned in 3 days. How model choice, context size, and Composer usage affect costs. Practical tier list and optimization strategies.