Prompt Cache

Prompt Cache is an inter-request cache at the API level: identical prefixes of different requests reuse the already computed KV Cache. Reading from the cache is significantly cheaper (approximately 10x cheaper for Anthropic and DeepSeek). Provider differences: Anthropic requires an explicit cache_control point, while OpenAI caching is automatic. Cache economics are not a post-hoc optimization, but an architectural constraint: the order of prompt elements is determined by the cache boundaries.

Related: KV Cache, [Three Rules of KV Cache Friendliness]