Caching

Opt-in result-cache layer for SERP and other read endpoints. Reduces latency and cost on hot queries. In development.

By default every SERP call hits the engine live for fresh results. The planned cache layer is opt-in and lets you trade absolute freshness for lower latency and cost on queries you make repeatedly inside a tight window.

The shape we're aiming for:

Per-endpoint TTL — each endpoint declares its sensible default cache window (e.g. 24h for SERP, much shorter for fast-moving surfaces).
x-ray9-cache: hit | miss response header so you can measure hit rate and reason about freshness.
Cache-Control: no-cache request header to bypass the cache for a single call without disabling it for others.
Reduced credit cost on cache hits versus live fetches.

Real-time stays the default — opt-in caching is for explicit performance and cost-sensitive workflows like rank trackers polling the same keyword set on a schedule.

In development — email contact@ray9.ai if it's blocking your roadmap.