AI SearchGEOLLM MentionsGoogle AI ModeAPI Guide

AI Search API: Track LLM Mentions, AI Visibility and Google AI Mode Answers

Four new SocQ AI Search endpoints return the AI Overviews and ChatGPT answers that cite a domain, AI share of voice, the most cited sources and full Google AI Mode answers.

SocQUpdated November 19, 20267 min read

If you work on generative engine optimization (GEO), you need data on what AI answers say: which questions cite your domain, how often your brand appears next to competitors, which sites the AI trusts for your category, and what Google's AI Mode writes for a search. On 2026-11-17 SocQ added four endpoints for this under the new AI Search platform. This guide covers what each returns, the request, the price and when to use it.

Quick answer: Send a domain (target) or a keyword (query) to POST /v1/ai-search/llm-mentions, /llm-brand-metrics or /llm-top-mentioned to read AI Overviews or ChatGPT answers. Send a search to POST /v1/ai-search/google-ai-mode for one AI Mode answer with its sources. Poll GET /v1/tasks/{task_id} for the results.

The New AI Search Endpoints

1 SocQ credit = $0.005.

EndpointWhat it returnsPrice
LLM Mentions APIOne row per AI answer that cites a domain or mentions a keyword: the question, the full answer in Markdown, cited sources and AI search volume40 credits ($0.20) per request + 0.5 credits ($0.0025) per answer
AI Search Visibility APIA total of AI answers that mention a domain or keyword, then breakdowns by cited domains, web-result domains and (on ChatGPT) brands and brand categories40 credits ($0.20) per request + 0.5 credits ($0.0025) per row
LLM Top Sources APIThe top ten domains or pages AI answers cite for a keyword or domain, ranked by mentions40 credits ($0.20) per request + 0.5 credits ($0.0025) per ranked domain or page
Google AI Mode APIThe full Google AI Mode answer for a search in Markdown, with every cited source2 credits ($0.01) per request

For the three LLM endpoints the 40-credit base also applies when nothing is found.

Authentication and the Async Task Flow

All SocQ requests authenticate with an API key in the Authorization header, and the same header is used for submit, task status and result requests:

Authorization: Bearer YOUR_API_KEY
Content-Type: application/json

Every collection runs as a background task:

submit -> task_id -> queued -> running -> succeeded
                                   \-> failed

The submit response contains data.task_id and data.status. Poll the task endpoint with that ID:

curl "https://api.socq.ai/v1/tasks/$TASK_ID?limit=50" \
  -H "Authorization: Bearer $SOCQ_API_KEY"

When status is succeeded, the response includes the first page of normalized records in data.results.items. limit defaults to 50 and goes up to 500. If has_more is true, pass results.next_cursor as cursor to read the next page. While a task is queued, running or failed, the endpoint returns task metadata without results.

An optional callback_url in the submit body makes SocQ send one POST with a task summary when the task finishes; read the records with GET /v1/tasks/{task_id}. A client-side timeout does not cancel a task, so resume with the same task ID instead of paying for a duplicate. 402 means insufficient credits; 429 means the key hit a rate or credit limit.

LLM Mentions API: The Answers That Cite You

The LLM Mentions API lists the questions where Google AI Overviews or ChatGPT cite your domain, or where your brand or keyword appears in the answer. Provide a domain in target or a keyword in query, not both. platform is google (AI Overviews, the default) or chat_gpt, which is tracked for the United States in English. results_limit defaults to 100 and accepts 1 to 1,000.

curl -X POST "https://api.socq.ai/v1/ai-search/llm-mentions" \
  -H "Authorization: Bearer $SOCQ_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "target": "notion.so",
    "platform": "google",
    "results_limit": 100
  }'

Each row is one question and answer. The fields you will use most:

  • question and answer: the question and the full answer in Markdown.
  • ai_search_volume: estimated monthly times the question is asked.
  • sources: cited pages with position, title, url, domain, source and text.
  • search_results and brands: web results the AI looked at and brands it named (ChatGPT only).
  • fan_out_queries: follow-up searches the AI ran.
  • first_seen_at and last_seen_at: when the answer was first and last recorded.

Use it to audit GEO coverage, find the pages AI trusts in your niche, or catch answers that describe your product wrongly. 100 answers cost 90 credits ($0.45). See the LLM Mentions docs for every field.

AI Search Visibility API: Share of Voice in AI Answers

The AI Search Visibility API is a share-of-voice report. The first row is the total number of AI answers that mention your domain or keyword and the monthly AI search volume behind them. After it come up to 10 rows for each breakdown: the domains the answers cite most, the domains in the web results the AI looked at and, on chat_gpt, the brands and brand categories named in the same answers.

curl -X POST "https://api.socq.ai/v1/ai-search/llm-brand-metrics" \
  -H "Authorization: Bearer $SOCQ_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "target": "hubspot.com",
    "platform": "chat_gpt"
  }'

Each row has a dimension (total, source_domain, search_result_domain, brand or brand_category), a name, a rank within its dimension, mentions_count and ai_search_volume. Run it for your brand and two competitors to compare who AI recommends. A ChatGPT report with the total and four full breakdowns is 41 rows, or 60.5 credits ($0.3025). Field reference: AI Search Visibility docs.

LLM Top Sources API: The Sites AI Quotes for a Topic

The LLM Top Sources API answers a different question: which sites does AI quote when people ask about your category? Give a keyword in query or a domain in target. Set content_type to domains (the default) or pages. results_limit defaults to 10, which is also the maximum.

curl -X POST "https://api.socq.ai/v1/ai-search/llm-top-mentioned" \
  -H "Authorization: Bearer $SOCQ_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "crm software",
    "platform": "google",
    "content_type": "domains"
  }'

Rows use the same metric shape as AI Search Visibility, with dimension set to domain or page. The page view is a ready shortlist for digital PR outreach. Ten ranked rows cost 45 credits ($0.225). Details: LLM Top Sources docs.

Google AI Mode API: One Full Answer With Sources

AI Mode is the conversational tab in Google Search that writes a complete answer instead of listing links. The Google AI Mode API takes a query and returns that answer as Markdown, plus a deduplicated list of the sources it cites. The defaults are the United States (location_code 2840) and English (language_code en); change them to compare answers across countries and languages.

curl -X POST "https://api.socq.ai/v1/ai-search/google-ai-mode" \
  -H "Authorization: Bearer $SOCQ_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "best crm for startups",
    "location_code": 2840,
    "language_code": "en"
  }'

Each request returns one row with query, answer, references (position, title, URL, domain, source and excerpt), location_code, language_code and the url of the AI Mode page. At 2 credits ($0.01) per search it is a low-cost way to check whether your pages are among the sources for the searches you care about. See the Google AI Mode docs.

Which Endpoint to Use

  • "Which AI answers mention us, and what do they say?" Use LLM Mentions.
  • "How visible are we in AI answers compared with competitors?" Use AI Search Visibility, once per domain.
  • "Which sites should we pitch so AI cites us?" Use LLM Top Sources with content_type: "pages".
  • "What does Google AI Mode say for this exact search, in this market?" Use Google AI Mode.

A common order is LLM Top Sources for the category, AI Search Visibility for you and your competitors, then LLM Mentions to read the answers behind the numbers. For classic rankings, see the SEO platform.

FAQ

What is an AI search API?

An API that returns what AI answer engines say instead of a list of links. SocQ's version covers AI Overviews and ChatGPT answers that mention a domain or keyword, visibility metrics, the most cited sources and full Google AI Mode answers.

Which AI platforms are covered?

The three LLM endpoints accept platform set to google (AI Overviews, the default) or chat_gpt. ChatGPT is tracked for the United States in English. Brand and brand-category rows, the AI's web results and brand lists appear only for chat_gpt.

Can I send a domain and a keyword in the same request?

No. LLM Mentions, AI Search Visibility and LLM Top Sources take either target (a domain) or query (a keyword), not both. Run two requests if you need both views.

Am I charged when no AI answer is found?

Yes, for the three LLM endpoints the 40-credit base applies when nothing is found; the per-result credits only apply to rows returned. Google AI Mode costs 2 credits per request.

How much does a typical GEO audit cost?

One LLM Top Sources request (45 credits), three ChatGPT visibility reports (up to 60.5 credits each) and one 100-answer LLM Mentions request (90 credits) come to at most 316.5 credits, about $1.58. Pricing can change, so check the endpoint pages first.

AI SEARCH API

Test the workflow with a public AI Search URL

Submit public inputs and receive normalized records with traceable source context.

Explore AI Search APIs