Caching & freshness

Identical queries are cached so you don’t pay full price (or wait) for a result that was just fetched. Whether a request was served from cache is always reported in stats.cached, and it determines the cost.

Cache hit vs fresh fetch

Source	`stats.cached`	Credits	Latency
Cache hit	`true`	3	sub-100ms
Fresh fetch	`false`	6	live fetch via residential proxy

A result is cacheable by the exact combination of query + country (gl) + language (hl) + time filter (tbs) + page — change any of them and it’s a different cache entry. See Pricing & credits for the credit model.

Controlling freshness per request

The optional freshness parameter lets you choose how a single request treats the cache:

`freshness`	Behaviour
`default`	Serve a recent cached result if one exists, otherwise fetch fresh. (default)
`realtime`	Always fetch fresh — bypass the cache. Charged the fresh rate (6 credits).
`cached`	Prefer a cached result.

{ "q": "nvidia stock price", "freshness": "realtime" }