Prompt Composition & Model Config
Tokens: ~1,280
Workload:
Expected Output 400 tokens
Anthropic Prompt Caching (1024+ Min) Injects cache_control: {"type": "ephemeral"} at system prompt breakpoint.
System Instructions & Tools Cached Block ~820 tokens
Context & Few-Shot Corpus ~340 tokens
User Query / Ephemeral Input ~120 tokens
<