> ## Documentation Index
> Fetch the complete documentation index at: https://docs.gladeapi.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Data Freshness and Caching

> Understand Glade API cache targets and response headers that identify cache status and source-data freshness.

Glade API caches normalized operation results. A request can therefore return cached data instead of making a new provider request. Cache hits and provider-backed results have the same response schema and each successful operation costs one unit.

## Cache targets

| Operation                                         | Cache target |
| ------------------------------------------------- | -----------: |
| Product details, sales estimates, seller profiles |      6 hours |
| ASIN/GTIN conversion                              |      30 days |
| Product variants                                  |     12 hours |
| Stock estimates, product search, best sellers     |   15 minutes |
| Product offers and deals                          |    5 minutes |
| Product reviews and autocomplete                  |       1 hour |
| Category products                                 |   30 minutes |
| Category taxonomies and author profiles           |     24 hours |

These are cache targets, not freshness guarantees. When a provider is temporarily unavailable, Glade API may serve a previously cached result rather than fail the request.

## Response headers

Use these headers to inspect the result:

| Header                    | Meaning                                                   |
| ------------------------- | --------------------------------------------------------- |
| `X-Glade-Cache`           | `hit`, `miss`, or `stale`                                 |
| `X-Glade-Data-Fetched-At` | ISO 8601 time when the underlying Amazon data was fetched |
| `X-Glade-Schema-Version`  | Version of the normalized response contract               |
| `X-Glade-Usage-Units`     | Units charged for this HTTP exchange                      |

For monitoring workflows, compare `X-Glade-Data-Fetched-At` rather than assuming that every request represents a new provider fetch.