LLM Mentions
An LLM mention is a question answered by Google AI Overviews or ChatGPT in which a domain is cited or a keyword appears, kept with the full answer, the pages it cites, the model behind it, how often the question is asked each month and, for ChatGPT, the brands named in the answer and the follow-up searches it ran. Reading those answers side by side shows which questions put a brand in front of AI users, which sources the models lean on and where an answer describes a product inaccurately, which feeds generative engine optimization audits, content planning and competitor comparison.
LLM Mentions API Features
- A domain or a keyword as the starting pointtarget looks for answers that cite a domain such as notion.so, while query looks for answers that mention a keyword or brand name, so you can track both your site and how people phrase your category.
- The answer exactly as users read itThe LLM Mentions API keeps the full response in answer, in Markdown with its headings, lists and inline links, so you can quote or review the wording the model gave.
- Every page the answer citessources lists each cited page with its position, title, URL, domain, site name and excerpt, which shows whose content shaped the answer.
- Demand behind each questionai_search_volume estimates how many times the question is asked per month, so mentions on high-demand questions can be told apart from rare ones.
- Brands, web results and follow-up searches on ChatGPTWith platform set to chat_gpt, brands lists the brands named in order, search_results shows the pages ChatGPT consulted and fan_out_queries holds the extra searches it ran.
- When an answer was recordedfirst_seen_at and last_seen_at date each answer, which helps separate long-standing answers from recent ones.
LLM Mentions API Parameters
| Parameters | Required | Description |
|---|---|---|
target | Required (at least one) | Domain to track in AI answers, such as hubspot.com, without a page path. Send either target or query. |
query | Required (at least one) | Keyword or brand name to track in AI answers, such as crm software, up to 250 characters. Send either target or query. |
platform | Optional | google reads Google AI Overviews and is the default; chat_gpt reads ChatGPT answers for the United States in English. |
location_code | Optional | Market location code. Defaults to 2840 (United States), the market covered by chat_gpt answers. |
location_name | Optional | Market location name, an alternative to location_code. When filled, it is used instead of location_code. |
language_code | Optional | Answer language code. Defaults to en, the language of chat_gpt answers. |
language | Optional | Answer language name, an alternative to language_code. When filled, it is used instead of language_code. |
results_limit | Optional | Maximum number of answers from 1 to 1000. Defaults to 100. |
How to Use the LLM Mentions API
Send a domain or a keyword to the LLM Mentions API and read one row per AI answer.
- POST {"target": "notion.so", "platform": "google", "results_limit": 100} to /v1/ai-search/llm-mentions.
- Poll GET /v1/tasks/{task_id} until status is succeeded.
- Read the answers from results.items and request the next page with results.next_cursor while has_more is true.
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}'LLM Mentions API Use Cases
- Auditing a domain's presence in AI answers: Run your domain as target in the LLM Mentions API, sort the rows by ai_search_volume and read answer and sources to see which high-demand questions cite you and how you are described.
- Finding the pages AI relies on in a category: Run a category keyword as query and count the domains and URLs in sources to build a list of articles worth pitching or improving on.
- Comparing brands inside ChatGPT answers: Use platform chat_gpt for your brand and competitors and compare brands, including each brand's position, across the same questions.
- Planning content from unanswered angles: Collect question and fan_out_queries from answers where your brand is missing to turn them into briefs for pages the models can cite.
LLM Mentions API Pro Tips
- Run both forms for one brand, the domain in target and the brand name in query, because citations of a site and mentions of a name return different sets of answers.
- Ask for the answers you need in one request with a higher results_limit, since the LLM Mentions API applies its 40-credit base to every request.
- Keep id from each LLM Mentions API row between runs and compare last_seen_at to tell new answers from ones you have already reviewed.
Related APIs
Beyond the LLM Mentions API, use these APIs when you need a different type of public AI Search data.
- AI Search Visibility API — Measure how often AI answers mention a domain or keyword, with AI search volume, the most cited domains and, on ChatGPT, the brands named alongside it.
- LLM Top Sources API — Rank the 10 domains or pages that AI Overviews and ChatGPT cite most for a keyword or domain, by mentions and AI search volume.
- Google AI Mode API — Get the full Google AI Mode answer for a search in Markdown, with every cited source, for any country and language.