Google SERP & AI Mode API — Search Results & Citations

At a glance

• Collect: Google SERP pages or AI Mode responses; optional deferred AI Overview completion.
• Input: use the runnable example below or the prefilled Input form.
• Output: the default dataset supports JSON/CSV/Excel export; see output modes below.
• Billing: $1.50 per 1,000 SERP pages/non-empty AI responses; $0.75 per 1,000 completed optional Overview follow-ups. No startup fee. Platform usage may be additional.
• Limits: AI answers/citations are not guaranteed. Rank absence means not found on fetched pages, not absent from Google.

Search Google in two modes from one Actor. Use SERP for organic rankings, ads, local packs, People Also Ask, related searches and available AI Overviews. Use AI Mode for Google's dedicated AI answer, source citations and available shopping cards.

No Google API key, login or proxy configuration is required. Batch keywords and export JSON, CSV or Excel, or call the Actor through the Apify API from n8n, Make or your own application.

Launch pricing — $1.50 per 1,000 main requests

Delivered output	Price per 1,000	Billing unit
Google SERP	$1.50	Successful search result page, including all returned sections
Google AI Mode	$1.50	Non-empty AI response, including text blocks and citations
Optional deferred AI Overview completion	$0.75	Completed non-empty follow-up response

No startup fee, subscription or per-link charge. Inline AI Overviews, domain-rank calculations, available ads, local packs and People Also Ask are included with the SERP page. Deferred completion is opt-in and charged separately only when content is delivered. Failed fetches and empty AI Mode responses have no Actor request fee; valid empty SERP pages remain billable. For example, 100 successful one-page searches cost $0.15 in Actor request fees. See the live Pricing tab for current rates and any applicable platform charges.

Quick start: standard Google search

{"mode":"serp","queries":["python tutorial"],"maxRequests":1,"gl":"us","hl":"en"}

Google AI Mode

{"mode":"ai_mode","queries":["python vs javascript for beginners"],"maxRequests":1,"gl":"us","hl":"en"}

AI Mode returns structured text_blocks, references, shopping_results and a convenient answerText. Google does not supply AI content for every query. An empty response produces a row with status: "unavailable", succeeds without an Actor request charge, and is counted separately in SUMMARY. AI Mode has no SERP pagination or strict language/country/time filters. Availability and citations vary between requests.

Domain rank tracking and pagination

{"mode":"serp","queries":["web scraping tools"],"targetDomains":["apify.com"],"maxPagesPerQuery":3,"maxRequests":3,"cache":false}

domainRanks reports matching organic URLs, page positions and bestRank for each fetched page. Rank is the starting offset plus the page position. Subdomains match their parent domain; unrelated suffixes do not. This is observed SERP position, not a universal or guaranteed ranking. Google redirect links that cannot be resolved may prevent a domain match. A missing domain is only absent from the fetched pages.

Pages are requested sequentially using offsets 0, 10, 20, up to 100. Desktop pagination stops when Google omits the next link or returns no organic results; mobile pagination uses offsets because next links are usually absent. maxRequests caps attempted requests, not result rows. Each dataset row contains one page or AI response, with its nested sections preserved. No automatic historical comparisons are promised.

Localization and filters

Set gl for the country perspective and hl for interface language. Optional location accepts Google's canonical names, for example New York,New York,United States; send a matching gl. uule takes precedence. Invalid or unsupported locations can be ignored by Google. Regional domains and mobile layout are supported.

SERP-only filters include strict language lr, strict country cr, time_period, spelling correction nfpr, and omitted-results filter. SafeSearch works in both modes. Leave SERP-only filters blank when using AI Mode. Fresh searches are the default; cached SERPs can be enabled with cache: true.

AI Overview versus AI Mode

An AI Overview is a panel within a standard search page; AI Mode is a separate answer surface. Inline AI Overviews are included with SERP output when available. A deferred panel is marked state: "deferred". Enable resolveAiOverview to attempt its completion immediately. This adds a separate follow-up request when available and counts toward maxRequests; set the limit accordingly. Completed follow-ups are separate mode: "ai_overview" rows. An unavailable follow-up is uncharged. Follow-up errors preserve the successful SERP and appear as warnings in SUMMARY.

Billing and output

See the live Pricing tab for active rates. SERP pages use serp-request; non-empty AI answers use ai-request; completed deferred panels use overview-request. There is no startup event. All sections delivered with a page are included; there is no per-link fee. Valid empty SERP pages are billable. Failed fetches and empty AI Mode responses are not charged by this Actor. Platform usage charges, if applicable, are separate. Set Apify's maximum total charge and maxRequests to control spending.

The dataset preserves structured sections, timestamps, locale and per-page domain ranks. SUMMARY contains attempted/successful/billable request counts, organic counts, unavailable AI responses, warnings and safe errors. Raw HTML and service session identifiers are not exposed. Fields can be empty or absent when Google does not supply them.

Support

Open an Actor issue with your run ID and non-sensitive input, or contact the developer through the profile for custom workflows and additional data sources. Never post credentials. Independent tool; not affiliated with or endorsed by Google. Results are point-in-time observations and AI answers can be inaccurate.

Minimal API example

Store your own Apify token in APIFY_TOKEN; never paste it into public input or code. This example uses the same small input as the prefilled form.

import os
import requests

response = requests.post(
    "https://api.apify.com/v2/acts/gauravlabs~google-serp-api/run-sync-get-dataset-items",
    headers={"Authorization": "Bearer " + os.environ["APIFY_TOKEN"]},
    json={"queries":["python tutorial"]},
    timeout=300,
)
response.raise_for_status()
print(response.json())

For large batches, start an asynchronous run with /runs, wait for its final status, and retrieve its default dataset. HTTP success when starting a run does not prove the extraction succeeded.

Minimal output shape (illustrative)

The following shows one documented field, not a fresh result or a promised yield. Inspect the output reference above for the complete schema.

{
  "mode": "serp"
}

Practical workflow and accuracy notes

SERP example: {"queries":["python tutorial"],"mode":"serp","maxRequests":1}. AI Mode example: {"queries":["how to choose a laptop"],"mode":"ai_mode","maxRequests":1}. Optional Overview follow-up example: {"queries":["what is machine learning"],"mode":"serp","resolveAiOverview":true,"maxRequests":2}. Overview/AI availability varies. Use the documented mode field and stop/status metadata to distinguish an absent answer from a request failure.