G

Google

Published byJoJ APIVerified publisher
9/14/2026
Agent payments

Endpoints

4

SERP
2

SERP (Search Engine Results Page) Endpoints

Tasks
2

Queue up to 100 searches at once and collect the results later, via webhook or polling. Built for volume jobs that don't need an instant answer.

What is the Google API?

Google Search APIs

Google API is available on the JoJ API marketplace. Subscribe with a single JoJ API key, call it through one gateway, and pay from one wallet — no separate account or contract with the provider.

Pricing & plans About this API Discussions More APIs by @jojapi Verified publisher OpenAPI spec MCP server

Google API endpoints

  • GET/v1/serp

    Full SERP

    One Google result page for a query. Every field comes off the HTML google.com served: `results` are the organic results in Google's order, each with the URL its own link resolves to (`matched: goto`) or the one Google printed in the clear (`matched: direct`); `ads`, `ai_overview` (with `text`, `html` and `references`), `people_also_ask`, `related_searches` and `page_blocks` (what sat where — an AI Overview above position four is the difference between "ranked fourth" and "fourth thing on the page"). `location_used` is where Google placed the search, always present. There is no `limit`: Google ignores `num`, a page shows what it shows, and depth is bought a page at a time with `page`.

  • GET/v1

    Basic SERP

    Get web search results, knowledge panel and related keywords

  • POST/v1/serp/tasks

    Queue Full SERP queries

    The same query as `/v1/serp`, queued. Submit up to 100 at once and get an id per task back at once (202); a worker runs them at the rate the pool sustains, and the answers are fetched with `/v1/serp/task` — each exactly what `/v1/serp` would have returned, or the error it would have answered with. Give a `callback_url` and a small notification (`{id, status, http_status}`) is POSTed to it as each task finishes, retried for hours if the endpoint does not answer 2xx; without one, poll `/v1/serp/task` with up to 100 ids per call. Every task is checked at submission with the same rules as `/v1/serp`, so a bad one refuses the whole submission (400 with `task_index`) and nothing is queued or billed. Billed per task queued; fetching is free. Results stay fetchable for 48 hours. This is the right entry point for batches: a night's keywords become queue depth rather than a wall of 429s, and the retries on the way are absorbed rather than surfaced.

  • GET/v1/serp/task

    Fetch queued Full SERP results

    The state and, once done, the answer of queued tasks. One id returns `task`; up to 100 comma-separated ids return `tasks` in the order asked, unknown ones as `not_found`. A done task carries `result` — exactly the `/v1/serp` answer — and a failed one `error` with the body and `http_status` `/v1/serp` would have answered. Free. Results are kept 48 hours.