AI search MCP and REST API for every model. One prompt in, every model's answer out.

Install Cite42 MCP in Claude Code, Codex CLI, Cursor, or Claude.ai, or call /v1/search over REST. One prompt fans out to the AI surfaces buyers use and returns each answer, citations, and structured data your agent can use.

  • $1 free on signup
  • Credits never expire
  • No subscription
  • MCP native
Get $1 free, try itInstall MCP

Step 1: Create your account

Get $1 in test credit to try your first calls.

Create free account

Step 2: Connect Cite42 MCP

Replace cite42_live_... with your key, then run this once.

Step 3: Ask inClaude Code

You ask your AIWhich customer-support automation tools should a Series A SaaS shortlist, and which sources do the models cite?
MCP runs/v1/search
Runs across the sources shown.
Your AI saysIntercom, Zendesk, Gorgias, and Help Scout appear across most AI answers. Perplexity cites G2 and vendor docs; Gemini leans on review roundups...
ENDPOINT PROMPTS

Ask for AI search data.

Cite42 calls /v1/search and returns structured data your AI can summarize.

  • Ask every model: best AI blog tool?

    You ask your AI“Ask every model: best AI blog tool?”
    MCP runs/v1/search
    Your AI saysPenflow shows up in the model answers. Claude and Perplexity cite its homepage; ChatGPT also mentions two competitors first. It includes the key result, source context, and next step...

  • What did Claude actually answer?

    You ask your AI“What did Claude actually answer?”
    MCP runs/v1/search
    Your AI saysClaude recommends Penflow for teams that want AI-assisted blog production, with 3 citations and one direct mention of your feature page. It includes the key result, source context, and next step...
RAW AI ANSWERS

Ask once, compare every answer.

Use AI search when your product needs the answer text, cited sources, and model-by-model differences before an agent writes or decides.

  • Compare consensus and outliers across AI surfaces.
  • Keep citations attached to the answer.
  • Store one normalized JSON shape.
WHAT WE QUERY

Sources Cite42 checks.

Cite42 queries AI answers, search demand, and conversation signals, then returns normalized results your app or agent can use.

  • ChatGPT

    OpenAI · Search when needed

  • Claude

    Anthropic · Search when needed

  • Perplexity

    Perplexity Sonar · Web-grounded

  • Gemini

    Google · Google Search grounding

  • Google AI Overviews

    Google Search · Live SERP

  • Google data

    Google data

    Search demand signal

  • Reddit

    Conversation signal

  • YouTube

    Video demand signal

EXAMPLE USE CASES

Three ways to use this endpoint.

A few practical ways teams use AI search.

  • Compare answers

    See where models agree.

    Run one category question and compare ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews side by side, with the answer text kept in one place.
  • Research agent

    Feed agents raw answers.

    Give an agent answer text, citations, and model differences before it writes a brief, memo, or recommendation.
  • Source check

    Audit cited domains.

    Find which sites are cited for a topic, which sources repeat across models, and which useful sources are missing.
PRICING

Pay per call for this endpoint.

Top up once ($25 minimum). Credits never expire, and this endpoint only charges when it runs.

EndpointPriceWhat you get
POST/v1/search$0.01-$0.07/modelFull prose answer + citations from each model, $0.01-$0.07/model depending on surface, all five for $0.22
See all endpoints and the source coverage table
KEEP EXPLORING

All endpoints

Same API key, same credit balance, different question.

QUESTIONS

Questions about AI search.

It returns one result per selected AI source: answer text, normalized citations, source URLs and titles, latency, and billing metadata. Use it when you need raw AI answers before calculating rankings, sentiment, or citation gaps.
ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews. Omit "models" for the full sweep, or pass a subset when you only need specific sources.
It is the closest API-equivalent answer. Cite42 uses provider APIs with web search or grounding enabled; consumer apps can still differ because vendors change UI prompts, personalization, and ranking behavior.
Use /v1/search when you want full answer text and citations. Use /v1/rankings or /v1/compare when you want scored brand visibility, mention rate, position, and share of voice.
Yes. Install the Cite42 MCP server or hosted connector, then ask in plain English. The agent calls the same /v1/search contract behind the scenes.
Every source comes back as a normalized array of { url, title } objects, even when the provider exposes sources differently. Your app or agent does not need per-provider citation parsing.
The API is stateless: no conversation history or memory. Cite42 stores billing records and response-cache entries so identical prompt and source calls can be served faster for a limited time.
READY WHEN YOU ARE

Give your AI live search answers.

Claude Code, Codex CLI, Cursor, or your app can pull current AI answers with one key.

Get $1 free, try itInstall MCP