Price alert Hurry up! Google SERP API from $0.03/1K calls — prices likely to rise 10× from December 2026.Hurry up! Google SERP from $0.03/1K — likely 10× from Dec 2026Google SERP $0.03/1K · likely 10× from Dec$0.03/1K · likely 10× in Dec Deposit Now
ChatGPT Claude Gemini Perplexity

AI Rank API — Brand Mentions (URL check)

Aggregate citations of your brand by AI LLMs.

Try for Free View Documentation
Per check Four AI engines One aggregate view Free trial calls

Call GET /api/ai/brand-mentions with one website address. The response is the buyer searches that qualified, each engine’s citations, and one aggregate view — as JSON. Part of the AI Rank API, inside Serpent AI Search Visibility.

Run a live brand mention check

One real check, no signup and no key — paste a website address and see the citations the four engines used.

One Endpoint, One Check, One Price

A check is one call: paste a website address and the work is billed once. The number of buyer searches, engines or citations never changes what the call costs.

$0.50 per check at Default, $0.05 at Growth, $0.025 at Scale, $0.020 at Enterprise and $0.015 at Enterprise Plus. Per 1,000 checks that is $500.00 at Default, $50.00 at Growth, $25.00 at Scale, $20.00 at Enterprise and $15.00 at Enterprise Plus. Growth requires one qualifying $100 deposit, Scale $500, Enterprise $1,000 and Enterprise Plus $1,250. New credit uses the current schedule; existing credit keeps its original schedule. Refunded or charged-back amounts do not qualify toward a tier. There is no monthly plan.

One check is one bill. max_queries runs from 1 to 5 and defaults to 5. The price does not change with it, so it is a question of how much of the answer you want in one call, not how to trim the bill.

Every account starts with free API calls, and AI Brand Mentions is free-trial eligible — it draws from the shared 10-call free pool, then bills per check. A check the service could not run is refunded; a completed check is charged.

What One Check Returns

One JSON object per check, the same shape whatever the engines answered. Every key is present on every successful response.

The website

  • success
  • brand.domain
  • brand.url

The buyer searches

  • queries[].phrase
  • queries[].brand_in_top_10
  • queries[].brand_position

Per engine

  • engines.chatgpt
  • engines.claude
  • engines.gemini
  • engines.perplexity
  • engines.<engine>.model
  • engines.<engine>.total_citations
  • engines.<engine>.target_found
  • engines.<engine>.error

Each citation

  • engines.<engine>.citations[].url
  • engines.<engine>.citations[].domain
  • engines.<engine>.citations[].title
  • engines.<engine>.citations[].position

The aggregate

  • aggregate.total_citations
  • aggregate.engines_citing[]
  • aggregate.unique_urls[]
  • aggregate.best_position

Run metadata

  • meta.elapsed
  • meta.timestamp
  • meta.max_queries

When the answer is short

  • meta.partialResults.requested
  • meta.partialResults.returned
  • meta.partialResults.reason
  • delivery — the same counts, in the shared shape
  • absent on a complete answer

Request parameters

  • url — required; a full URL or a domain
  • domain — accepted alias for url
  • country — optional two-letter code
  • language — optional two-letter code
  • max_queries — 1 to 5, default 5
  • X-API-Key — request header

Four Engines, One Response Shape

ChatGPT, Claude, Gemini and Perplexity each answer the same buyer searches. Each one’s citations arrive under its own key, in the same shape, so one parser covers the whole set.

ChatGPT
gpt-6.1-sol
  • The citations its answers used when they mentioned the brand
  • total_citations and target_found for the check
  • error is null when the engine answered
Claude
claude-sonnet-5-5
  • The citations its answers used when they mentioned the brand
  • total_citations and target_found for the check
  • error is null when the engine answered
Gemini
gemini-4-argon
  • The citations its answers used when they mentioned the brand
  • total_citations and target_found for the check
  • error is null when the engine answered
Perplexity
sonar
  • The citations its answers used when they mentioned the brand
  • total_citations and target_found for the check
  • error is null when the engine answered

Every response carries a model field with that engine’s public model label. Store it beside each result: when a label changes, a shift in your chart has an explanation sitting in your own data instead of looking like unexplained movement.

Which Buyer Searches Qualified

You send a website, not a keyword. The check works out the short searches buyers would type about the business, then keeps the ones where the brand already appears in the top 10 results.

phrase

The short search a buyer would type about this kind of business. It is written for the website you sent, not picked from a keyword list.

brand_in_top_10

true when the brand already appears in the top 10 results for that search. Only the searches where it does are put to the four engines.

brand_position

Where the brand appeared in that top 10, with 1 as the first result. It is null when the brand did not appear.

max_queries

How many buyer searches one check works through: 1 to 5, five by default. Every search that was checked appears in queries[], including the ones that did not qualify.

One Aggregate View per Check

The aggregate is the line to chart: one view across all four engines, derived from the per-engine rows beside it.

total_citations

Every citation from the four engines in one number.

engines_citing[]

The engines whose answers cited your brand’s own domain, always in the same fixed order: chatgpt, claude, gemini, perplexity.

unique_urls[]

Every cited URL once, deduplicated and sorted, so a page cited by two engines is not double-counted in your own reporting. They are normalised for comparison — the scheme and a www. prefix are dropped — so a page does not appear twice under two spellings.

best_position

The best position the brand held in any engine’s answer, with 1 as the first source named; null when no engine mentioned the brand.

Under-delivered is still delivered. If the check runs out of time, you get the citations gathered so far. meta.partialResults records what was requested and what was returned, and a delivery block states what was handed over. The response shape does not change, so your parser never has to test for a missing key.

One Request, One Website Address

url is the only required parameter. country, language and max_queries are optional.

cURL — one brand mention check
# One website address is enough. Add country, language or max_queries if you want them.
curl "https://api.apiserpent.com/api/ai/brand-mentions?url=https://example.com" \
  -H "X-API-Key: YOUR_API_KEY"
Python — requests
# pip install requests
import requests

resp = requests.get(
    "https://api.apiserpent.com/api/ai/brand-mentions",
    params={"url": "https://example.com", "country": "us", "language": "en", "max_queries": 5},
    headers={"X-API-Key": "YOUR_API_KEY"},
    timeout=90,
)
data = resp.json()

print("brand:", data["brand"]["domain"])
print("citations:", data["aggregate"]["total_citations"])
print("cited by:", data["aggregate"]["engines_citing"])
for name, row in data["engines"].items():
    print(name, row["total_citations"], row["target_found"], row["error"])
JSON Response (abridged)
{
  "success": true,
  "brand": { "domain": "example.com", "url": "https://example.com/" },
  "queries": [
    { "phrase": "best project management tools", "brand_in_top_10": true, "brand_position": 4 },
    { "phrase": "project management software for small teams", "brand_in_top_10": true, "brand_position": 7 },
    { "phrase": "example project management platform", "brand_in_top_10": false, "brand_position": null }
  ],
  "engines": {
    "chatgpt": {
      "model": "gpt-6.1-sol",
      "citations": [
        { "url": "https://asana.com/best-project-management-tools", "domain": "asana.com", "title": "Best project management tools in 2026", "position": 1 },
        { "url": "https://example.com/compare", "domain": "example.com", "title": "Project management comparison", "position": 2 }
      ],
      "total_citations": 2,
      "target_found": true,
      "error": null
    },
    "claude": {
      "model": "claude-sonnet-5-5",
      "citations": [
        { "url": "https://example.com/compare", "domain": "example.com", "title": "Project management comparison", "position": 3 }
      ],
      "total_citations": 1,
      "target_found": true,
      "error": null
    },
    "gemini": {
      "model": "gemini-4-argon",
      "citations": [],
      "total_citations": 0,
      "target_found": false,
      "error": null
    },
    "perplexity": {
      "model": "sonar",
      "citations": [
        { "url": "https://example.com/compare", "domain": "example.com", "title": "Project management comparison", "position": 1 },
        { "url": "https://monday.com/project-management", "domain": "monday.com", "title": "Project management software", "position": 4 }
      ],
      "total_citations": 2,
      "target_found": true,
      "error": null
    }
  },
  "aggregate": {
    "total_citations": 5,
    "engines_citing": ["chatgpt", "claude", "perplexity"],
    "unique_urls": ["asana.com/best-project-management-tools", "example.com/compare", "monday.com/project-management"],
    "best_position": 1
  },
  "meta": { "elapsed": "33891ms", "timestamp": "2026-11-19T09:12:44.000Z", "max_queries": 3 }
}

A check can take up to about 45 seconds, so set a client timeout above that rather than the 10 or 30 seconds you would use for a fast endpoint. Errors come back in the standard taxonomy: invalid_parameter (400), missing_api_key / invalid_api_key (401), insufficient_credits (402), rate_limit_exceeded (429), request_timeout (504), request_failed (502) and internal_error (500). A genuine empty answer is not an error — it is a normal 200.

What Teams Use a Brand Mention Check For

Four jobs the endpoint gets bought for, and the field each one reads.

A brand-level AI visibility report

One check gives a client-ready answer: which buyer searches the brand qualified for, which engines mentioned it, and the exact sources those engines used. aggregate.total_citations is the number for the summary line.

Content and PR proof

Save aggregate.unique_urls before a launch and again after it. New sources appearing in engines’ answers is evidence that a change travelled, held in your own storage rather than a vendor’s.

Competitive context

Every citation carries its domain, so the list shows which other sites the engines leaned on in the same answers as your brand. That is the source set around your brand, not just your own URL count.

Screen a site before AI Rank

The brand mentions check is per website and aggregated; the keyword endpoints are per keyword and per engine. Use this for the brand-level picture, then the AI Rank API for the keyword you want to chart every week.

Per Check, by Deposit Tier

A check is billed once per call. max_queries changes how much the check does, never what it costs.

Per check

  • Default — $0.50 per check
  • Growth — $0.05 per check
  • Scale — $0.025 per check
  • Enterprise — $0.020 per check
  • Enterprise Plus — $0.015 per check

Per 1,000 checks

  • Default — $500.00/1K checks
  • Growth — $50.00/1K checks
  • Scale — $25.00/1K checks
  • Enterprise — $20.00/1K checks
  • Enterprise Plus — $15.00/1K checks

The five tiers are set by deposit: Growth requires one qualifying $100 deposit, Scale $500, Enterprise $1,000 and Enterprise Plus $1,250. New credit uses the current schedule; existing credit keeps its original schedule. Refunded or charged-back amounts do not qualify toward a tier. There is no monthly plan, and no minimum spend beyond the deposit that sets your tier.

The free trial works like every other Serpent API: new accounts get free API calls, and AI Brand Mentions is eligible to spend them. After that, each check bills at your tier’s per-check rate. A check the service could not run is refunded; a completed check is charged even when it found no citations.

The tier is the only lever on the per-check price. At Enterprise Plus a check is $0.015 and 1,000 checks are $15.00, against $0.50 and $500.00 at Default. Because a check is billed once whatever max_queries is, buying a higher tier is the only way to lower what a check costs.

AI Brand Mentions Questions

The brand mentions check is the URL-based endpoint of the AI Rank API. It turns one website address into the citations AI engines used when their answers mentioned that brand. Send a full URL or a domain: the service works out the short searches a buyer would type about the business, keeps the ones where the brand already appears in the top 10 results, then asks the four AI engines those searches. The response is every citation they used in answers that mentioned the brand, per engine plus one aggregate view. In one line: aggregate citations of your brand by AI LLMs.
A check can take up to about 45 seconds. Once the first buyer searches have been checked, a timeout still returns what was gathered: meta.partialResults records what was requested and what was returned, and a delivery block states what was handed over. If nothing had been checked yet, the call answers the standard 504 request_timeout instead. On a successful answer the shape never changes, so your parser never sees a missing key.
Per check, by deposit tier: $0.50 on Default, $0.05 on Growth, $0.025 on Scale, $0.020 on Enterprise and $0.015 on Enterprise Plus. Per 1,000 checks that is $500.00, $50.00, $25.00, $20.00 and $15.00. A check is billed once per call, whatever max_queries is. AI Brand Mentions is free-trial eligible: it draws from the shared 10-call free pool, then bills per check. Growth requires one qualifying $100 deposit, Scale $500, Enterprise $1,000 and Enterprise Plus $1,250. There is no monthly plan.
Only when the service itself could not run the check, for example when no buyer searches could be worked out or every engine failed on every qualifying search. A check that ran to a real answer is charged even when it finds nothing: a brand that genuinely does not appear in the top 10 for a buyer search is an answer, not a failure. Refunds are automatic and are not a field in the response.
Yes, for any public website. Send a full URL or just a domain such as example.com. Country and language are optional two-letter hints, and a site with very little readable text may produce fewer buyer searches, but the check still returns a complete response for what it could check.
The keyword endpoints start with a keyword you choose and return the citations each of the four engines used for that one question. The brand mentions check starts with a website: it finds the searches buyers would type about that business, keeps the ones where the brand already appears in the top 10, and returns the citations the engines used in answers that mentioned the brand, with one aggregate view. The keyword endpoints are the per-keyword tracking tool; this endpoint is the per-website, aggregated view.
A citation is a URL an engine named as a source in its answer. Each one arrives with its url, domain, page title and position in that answer. The citations here come from answers that mentioned your brand, so the list shows the sources the engine leaned on in those answers, including third-party pages and competitor sites, not only your own. aggregate.unique_urls is the deduplicated set across the four engines.
Up to 5, and five by default. Set max_queries from 1 to 5. The price does not change, so it is a question of how much of the answer you want in one call, not how to trim the bill. Every search that was checked appears in queries[] with its brand_in_top_10 flag and brand_position, including the searches the brand did not appear in.

See what the AI engines cite when they mention your brand

Free API calls to start, no card and no subscription. One check is $0.50 at Default ($500.00 per 1,000 checks).

Try for Free

Related guides

Deep-dives on AI search visibility and answer-engine optimization.

What Does an AI Rank API Citation Actually Prove?Audit returned AI Rank citation URLs, cited text and domain matches before claiming brand visibility. Separate citation evidence from mentions and recommendations. AI Search Optimization GuideThe complete 2026 playbook for getting your brand into AI answers. How AI Picks CitationsWhy AI engines cite some sites and ignore others — and how to get picked.