Prompt Caching Playground

Optimize your prompt structure and observe cache hits. Cache engages at ≥ 1024 tokens, with cacheable chunks in 128-token increments (e.g., 1024, 1152, 1280…).

Setup

Use repeats to push your prompt over 1024 tokens (and observe cached increments).

Prompt Parts

Tip: Try A → A2 (same prefix, new question) to see cache hits; then A → B to see instruction-only prefix hits.

Result

cache status latency timestamp
Prompt Tokens
–
Cached Tokens
–
Cache Hit %
–
Completion Tokens
–
Total Tokens
–
Model
–
Endpoint
–
prompt_cache_key
–
Response will appear here…

Raw JSON

Most recent request JSON 
—
Most recent response JSON 
—

Run Log