Google data API · Search

Google Search API. Pay per record.

Run a Google web search and get the organic results as JSON: rank, title, link, snippet, domain and date, with related questions and searches. Operators like site:, intitle: and OR work, and you can filter by country, language, site, date range and file type.

Price
$0.0009 / page
Billed
Per record
Free to start
$0.10
Endpoint
POST /v1/google

Updated 21 October 2026

What people build with the Google Search API

Ground an agent in the web

Let an agent search before it answers, then read the links it found, with the rank of each.

Rank tracking

Check where a domain ranks for a set of queries in each country, on a schedule you choose.

Find evidence first

Search a site, such as one domain or a profile path, to find the exact page or handle before you look it up elsewhere.

Quote it, then run it

1. free quote: the exact price, nothing charged
curl -X POST https://apiforagents.dev/v1/google \
  -H "Authorization: Bearer $AFA_KEY" \
  -H "Content-Type: application/json" \
  -d '{"op":"search","query":"best CRM for startups","pages":1,"country":"US","language":"en","estimateOnly":true}'
2. the same call, run
curl -X POST https://apiforagents.dev/v1/google \
  -H "Authorization: Bearer $AFA_KEY" \
  -H "Content-Type: application/json" \
  -d '{"op":"search","query":"best CRM for startups","pages":1,"country":"US","language":"en"}'

Inputs

Everything you can send in one search call. Required inputs first.

Show all 9 rows
query *string≤300 chars
pagesintegerBilled per page of ~10 results. The one op counted in pages. 1-5
countrystringCountry to search from. ISO alpha-2
languagestringBCP-47
sitestringOnly results from this domain ("linkedin.com/in", "reddit.com"). The same as writing site: in the query. ≤120 chars
relatedToSitestringSites Google considers similar to this domain, instead of the query's own matches. ≤120 chars
sincestringYYYY-MM-DD
untilstringYYYY-MM-DD
fileTypestext listOnly documents of these types, as extensions ("pdf", "xlsx", "ppt"). Empty = any page. each ≤8 chars

What comes back

The fields of every row, always present. Nothing is guessed: a value that could not be measured is null, and the answer's coverage says why.

Show all 12 rows
linkstring
titlestring
snippetstring
domainstring
sourcestring
publishedDatestring
rankinteger
relatedQuestionsany
relatedSearchesany
aiAnswerstring
aiAnswerSourcesany
detailsobject

details is always the last field: everything else the source sent for that row, under our own field names. Its keys depend on the source that answered, so read them defensively; it is {} when there is nothing more.

A sample answer

Made-up names and values in the real field names, so you can see the shape before you spend anything.

1How to price a B2B SaaS product (2026 guide)northfold.exampleA step-by-step way to set your first price, with worked e…2026-08-12https://www.google.com/how-to-price-a-b2b-saas-product-20…
2SaaS pricing models comparedledgerly.examplePer seat, usage-based and flat fees, side by side.2026-06-30https://www.google.com/saas-pricing-models-compared-1001
3Pricing page examples that convertbrightdesk.example24 real pricing pages and what they get right.2026-09-18https://www.google.com/pricing-page-examples-that-convert…

Pricing

Pay as you go, in US dollars, per record. No subscription, nothing expires, $5 first top-up, and every new agent starts with $0.10 free.

One page1$0.0009

These are per-record list prices. Some calls also carry a minimum, such as a page of search results, so a small request can cost more than the table shows. The quote ("estimateOnly": true) is free, gives the most your exact request can cost, and the charge then settles to what came back. Records that do not come back are refunded.

Try it without writing code

Why this one, and not a scraper you run yourself

  • No login, no Google account, no cookies, no proxies, no browser to babysit.
  • When the site changes its pages, your request stays the same: keeping the reads working is our job, not yours.
  • A fetch that fails costs nothing: it is named in the answer with a reason, and the quote says the most a call can cost before it runs.

Use it from your tools

Questions people ask.

What does the Google Search API return?

One row per organic result: rank, title, domain, snippet, published date where shown, and link, plus the related questions and related searches Google shows for the query.

What do I pass in?

A query of up to 300 characters, and optionally the number of pages (one to five), country, language, a site to search within, a site to find similar sites to, since and until dates, and file types such as pdf or xlsx.

How is a search billed?

$0.0009 per page of about ten results at the list price. A small call can carry a minimum, so quote first with "estimateOnly": true: the quote is free and is the most the call can cost. You pay for what comes back.

Try it in a minute. $0.10 free.

The playground runs any endpoint by hand: pick a platform and what you want from it, fill a short form named in plain words (or press Use sample input), see the free quote on the Run button, run it, then read the answer as a table or JSON and download it as CSV or JSON. The same call is written out as cURL, Python and Node while you type.

Quickstart · Every price · All platforms

Why people pick API for Agents

  • One key for 26 platforms: social, commerce, jobs, places, search and contacts, with one request shape and one response shape.
  • A free quote before any spend: add "estimateOnly": true to any call and see the most it can cost.
  • Pay per record in US dollars. No subscription, nothing expires, top-ups from $5.
  • $0.10 free to try, no card. An AI agent can sign itself up in one request; a person claims it later.
  • Measured, not estimated: every answer says when each row was measured, whether it was fresh or a recent capture, and how sure to be of each field.
  • Partial success is success: anything that could not be fetched is named with a reason, and the charge settles to what came back, never above the quote.
  • Typed fields that keep their names whichever source answered, and a details object last with everything else the source sent, under our own field names.
  • Try any endpoint in the playground and download the answer as CSV or JSON.

API for Agents is independent and not affiliated with or endorsed by LinkedIn or any company named here. LinkedIn is a trademark of LinkedIn Corporation; other names belong to their owners. Public data only: nothing behind a login.