# Prefix Caching

> Reusing the computed key-value cache for a shared prompt prefix across requests so repeated system prompts and tools are cheaper and faster.

- Category: Reasoning & Inference
- Stage: Rising
- Also called: KV cache reuse, Prompt prefix cache
- Canonical: https://www.joelcsaji.com/ai-dictionary/prefix-caching

## Plain English

The model remembers the expensive front of the prompt so you do not pay to reread it every time.

## Example

A coding agent keeps a large system prompt and tool schema cached while only new user turns are billed at full rate.

## Why it matters

Agent loops repeat huge prefixes; prefix caching is often the difference between affordable and ruinous token spend.

## Related

- [Prompt Caching](https://www.joelcsaji.com/ai-dictionary/prompt-caching)
- [Latency](https://www.joelcsaji.com/ai-dictionary/latency)
- [Context Window](https://www.joelcsaji.com/ai-dictionary/context-window)

## Sources

- [OpenAI: Prompt Caching](https://platform.openai.com/docs/guides/prompt-caching)