Sign in Get API key
LLM Answers API

ChatGPT Answers with Cited Sources, Without an Account

Scrappa's LLM Answers API sends a prompt to the public, logged-out ChatGPT and returns the answer text, every cited source with link and snippet, citation positions, and the model that answered. It also lists live model catalogs, the public Perplexity answer feed, trending topics, and the public prompt library. Scrappa is an independent service and is not affiliated with or endorsed by OpenAI or Perplexity.

500 free credits
No card needed
From $0.20 / 1k
Pay as you go
Structured JSON
Ready for apps
API key access
Test in minutes
Start for free View documentation
Browse endpoints Try first endpoint
5
Endpoints
2
Answer engines
0
Accounts needed

Try it now

Make one request. Inspect structured JSON.

Use the first endpoint in this API family with a Scrappa API key. Every account includes 500 free monthly credits, so you can validate the response shape before buying credits.

Get free API key Try in Playground Open docs
Request preview
Request 
curl -X GET "https://scrappa.co/api/llm-answers/answer?prompt=What+is+the+tallest+mountain+in+Europe%3F" \
  -H "Accept: application/json" \
  -H "X-API-KEY: YOUR_API_KEY"
JSON response 
{
  "success": true,
  "data": {
    "answer": { "text": "The tallest mountain in Europe is Mount Elbrus, at 5,642 metres.", "length": 66 },
    "sources": [
      {
        "ref_id": "turn0search0",
        "title": "El'brus",
        "url": "https://education.nationalgeographic.org/resource/elbrus/",
        "attribution": "education.nationalgeographic.org",
        "cited": true
      }
    ],
    "citations": [{ "type": "grouped_webpages", "refs": ["turn0search0"], "start_idx": 66, "end_idx": 66 }],
    "search_queries": ["tallest mountain in Europe"],
    "model": { "requested": "auto", "resolved": "gpt-6" },
    "follow_up": { "available": false, "reason": "Follow-up suggestion text is not available." }
  },
  "meta": { "duration_ms": 7800, "upstream_calls": 3, "model_calls": 1 }
}

What you can build

Structured data for real applications. Here are some ideas.

AI Search Visibility

Check which sources and brands a public answer engine cites for the questions your customers ask.

Answer Monitoring

Track how the answer to a fixed prompt changes over time, including its cited sources.

Research Assistants

Give an agent a cited answer with source links instead of an unsourced paragraph.

Content Gap Analysis

Find the questions where competitors are cited and your pages are not.

Prompt Research

Browse trending topics and the public prompt library to find prompts worth tracking.

LLM Answers API

LLM Answers API for ChatGPT answers, cited sources, and Perplexity feed data

The LLM Answers API is built for AI search visibility work, answer monitoring, and research tools. Send a prompt and get the clean answer text, each cited source with title, link, snippet, and publish date, the position of every citation in the text, the web searches the model ran, and the model that answered. Follow-up suggestion text is not available, so the response says so instead of inventing it.

Ask a question and get answer.text plus sources, citations, search_queries, and model.resolved.

Ask a follow-up by sending the earlier turns in the history parameter.

List the live model catalog of ChatGPT or Perplexity and pass a model id to the answer endpoint.

Browse public Perplexity answers with their cited sources, page by page, by topic.

Read trending topics and the public prompt library for prompt ideas.

Failed requests are never billed. Test the LLM Answers API with 500 free monthly credits.

Choose the right LLM Answers API endpoint

Use Answer for cited answers

Send a prompt and read sources[] for the pages the answer cites. Each source has a ref_id that citations[].refs points to, so you can see which sentence relies on which page. Answers take 4 to 60 seconds.

Open LLM Answers Answer API docs

Use Models to pick a model

Read the live catalog instead of a fixed list, then pass a model id to the answer endpoint. The answer reports the model that really answered in model.resolved.

Open LLM Answers Models API docs

Use the Perplexity Feed for public answers

The feed returns finished public answers with their cited sources. It does not answer a prompt you submit. Page with limit and offset and filter with a topic id.

Open LLM Answers Perplexity Feed API docs

Available endpoints

Everything you need in one API.

Related resources

Google Search API overview

Compare the sources an answer engine cites with the pages that rank in Google search results.

Brave Search API overview

Add a second web search index to check which pages are visible beyond Google.

Scrappa vs ScrapingBee / ScraperAPI

See how Scrappa compares to the official API.

Answer text and cited sources as JSON
Scrappa:
ScrapingBee / ScraperAPI: Build your own parser
No account or login to manage
Scrappa:
ScrapingBee / ScraperAPI: Bring your own session
Citation positions in the answer text
Scrappa:
ScrapingBee / ScraperAPI:
Live model catalogs
Scrappa:
ScrapingBee / ScraperAPI:
Unified API key
Scrappa: Yes (80+ APIs)
ScrapingBee / ScraperAPI: Scraping only
Free tier
Scrappa: 500 credits/month
ScrapingBee / ScraperAPI: Trial credits

Key features

Clean answer text with citation markers resolved and citation positions mapped onto the text
Cited sources with title, link, snippet, publish date, and a cited flag
The web searches the model ran and the model that answered
Follow-up questions through the history parameter
Live model catalogs for ChatGPT and Perplexity
Public Perplexity answer feed with cited sources, paging, and topics
Failed requests are never billed
Simple API key auth with 500 free credits per month

Frequently asked questions

No. Scrappa is an independent service and is not affiliated with or endorsed by OpenAI or Perplexity. The LLM Answers API returns what the public, logged-out experience shows, as structured JSON.

No. The public stream only says whether suggestions exist and never sends their text. The response sets follow_up.available to false and does not invent suggestions.

The service keeps no conversation between requests. Send the earlier turns in the history parameter, for example history[0][role]=user&history[0][text]=..., then the new question in prompt.

No. The Perplexity endpoint returns existing public answers from the Discover feed. Prompts you submit are answered by ChatGPT.

Most answers take 4 to 15 seconds and some take up to a minute, because the model searches the web first. Set your client timeout to at least 90 seconds.

Only requests that return an answer. Requests that fail, return no answer text, or end early use no credits.

Every Scrappa account includes 500 free credits each month. After that, each request costs one credit.

Start using the LLM Answers API

Sign up and get 500 free credits. No credit card required.

Create Free Account View Documentation