The AI visibility API
for every AI search engine.
Get what ChatGPT, Gemini, Perplexity and Google AI actually show users: the full answer, every cited source and the ads, as structured JSON.
No card required · Only successful requests are billed
RequestPOST /v1/wire/task
curl anakin.io/v1/wire/task -H "X-API-Key: $KEY" -d '{
"action_id": "chatgpt",
"params": { "prompt": "best CRM for small teams?", "country": "us" }
}'The answer engines · live now
Every engine your customers track, unpacked.
Prompt in, structured answer out. Use it as a ChatGPT scraper API, a Perplexity or Google AI Overview API, or all of them under one key. Engines marked cites sources return the pages behind each answer.
The answer, its citations, fan-out queries and the ads ChatGPT shows.
Structured chat answers with grounding sources.
AI answer engine, returns the answer with citations and web sources.
The AI Overview shown above organic results.
Google's cited AI answers, plus Scholar search.
Naver AI Briefing, the AI answer on Korea's top search portal.
Baidu's AI assistant, China's leading answer engine.
API access
One call in, the whole answer out.
Send a prompt to any engine and get the full answer back as structured data: answer text, cited sources, query fan-out, shopping blocks and, for ChatGPT, the ads shown.
Request example
curl https://anakin.io/v1/wire/task \
-X POST \
-H "X-API-Key: $ANAKIN_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"action_id": "chatgpt",
"params": {
"prompt": "Best sneakers to buy under 300 dollars",
"web_search": true,
"include_html": false,
"country": "us"
}
}'X-API-Key header.Get API keyResponse example
{
"status": "ok",
"data": {
"prompt": "Best sneakers to buy under 300 dollars",
"title": "Sneaker recommendations under 300 dollars",
"answer_text": "If you mean sneakers broadly (running + everyday + basketball), you don't need to spend the full $300. Current 2026 reviews show strong options well below that ceiling…",
"answer_text_markdown": "If you mean **sneakers broadly**…\n\n### My shortlist\n- **adidas Hyperboost Edge: $200**\n- **Nike Dunk Low: $130**\n- **Nike G.T. Cut 4: ~$116**…",
"answer_html": null,
"citations": [
{
"url": "https://www.wired.com/gallery/best-running-shoes/",
"title": "Best Running Shoes, Tested and Reviewed (2026) | WIRED",
"domain": "Wired",
"date_published": "2026-06-05T00:00:00Z",
"ref": "turn0search1",
"ref_type": "search",
"cited": true,
"position": 1
},
{
"url": "https://runrepeat.com/guides/best-cheap-running-shoes",
"title": "7 Best Cheap Running Shoes in 2026 | RunRepeat",
"domain": "Runrepeat",
"date_published": "2026-09-01T00:00:00Z",
"ref": "turn0search2",
"ref_type": "search",
"cited": true,
"position": 3
}
],
"search_sources": [
{ "url": "https://www.wired.com/gallery/best-running-shoes/", "title": "Best Running Shoes (2026) | WIRED", "attribution": "wired.com", "rank": 1 }
],
"shopping": [
{
"title": "adidas Hyperboost Edge",
"price": "$200.00",
"link": "https://www.nordstrom.com/s/hyperboost-edge-running-shoe/8703559",
"purchasing_options": [
{ "merchant": "Nordstrom", "price": "$200.00", "available": true, "tag": "Best price" }
]
},
{
"title": "Nike G.T. Cut 4",
"price": "$115.97",
"rating": 4.7,
"link": "https://www.nike.com/t/gt-cut-4-…/IU1960-001",
"purchasing_options": [
{ "merchant": "Nike", "price": "$115.97", "original_price": "$210.00", "available": true }
]
}
],
"shopping_visible": true,
"web_search_triggered": true,
"web_search_query": "best sneakers under $300 2026 running lifestyle Nike Adidas New Balance",
"model": "gpt-5-6",
"country": "us"
},
"files": [],
"error": null,
"meta": {
"action_id": "chatgpt",
"catalog_slug": "ai-assistants",
"execution_ms": 36316,
"envelope_version": "v1"
}
}Run this on your own prompts. 300 free credits, no card.
Get 300 free creditsWho it's for
Built for teams shipping AI visibility.
GEO & AEO platforms
Your product runs on AI answers. Stop maintaining the scrapers behind it.
SEO & GEO agencies
One visibility audit for every client, on every engine, in a report they get.
AI product teams
Know what ChatGPT tells people about your product before users quote it back to you.
Brand & growth teams
See which sources AI engines trust, then go earn a spot in them.
Pricing
Buy the volume you need, pay less as you scale.
Start free with 300 credits. Paid plans from $100/mo, down to $0.90 per 1K at volume. That's about $1 per 1,000 engine answers.
Try every engine with full answers and citations.
Prototype and ship on every engine.
Steady tracking across brands and prompts.
A lower rate for teams tracking at scale.
Our best rate, custom terms and dedicated support.
Need a different volume?
Slide to any monthly volume from $100 to $4,500. The rate drops automatically as you go up.
Answers / month
Per 1,000 answers
$1.00Enough for about 500,000 engine answers a month, across any mix of ChatGPT, Gemini, Perplexity and the rest.
Your plan
Monthly500,000 answers, refilled monthly
- 500,000 answers every month
- $1.00 per 1,000 answers
- Every AI engine on one API key
- Change or cancel anytime
See the full volume pricing tableFull volume pricing table
| Purchase ($) | Price / 1K | Answers / mo | Concurrency |
|---|---|---|---|
| $100 – $900 | $1.00 | 100K – 900K | 25 – 100 |
| $1,000 – $2,500 | $0.95 | ~1.05M – 2.63M | 200 – 350 |
| $3,000 – $4,500 | $0.90 | ~3.33M – 5.00M | 400 – 550 |
Need more than $4,500/mo, or custom terms and volume?Talk to sales | |||
FAQ
Frequently asked questions
Everything prospects ask us, engines, output, reliability, and pricing, so you can evaluate without a single email.
Engines & Coverage
Do you capture ChatGPT ads?
Yes. Turn ads on in the request and the sponsored placements ChatGPT shows come back as structured data, alongside the answer and its citations. Ads data is priced separately from standard answers, so talk to us for rates at your volume.
Can I use this as a ChatGPT scraper API?
Yes. Call the chatgpt action and you get ChatGPT's full answer, its citations, fan-out queries and, if you ask for them, the ads, as structured JSON. The same key covers Gemini, Perplexity and Google AI.
Which AI engines do you support?
ChatGPT, Gemini, Perplexity, Google AI Overviews, Google AI Mode, Naver, and Wenxin (ERNIE). Google SERPs too.
Does it work on Google?
Yes. Google AI Overviews, AI Mode and the underlying Google SERP, each with answer text and citations, from any country you choose.
Data & Output
What do I get back?
The full answer text plus structured citations (title, URL, position), and the raw SSE stream as the source of truth. One clean, parseable object per request.
Is it the same as the real UI answer, or the official API?
Yes. It's pulled from the real web experience, so it's 1:1 with what a user actually sees in the app, including web search citations, shopping results and, on ChatGPT, ads. The official model APIs return raw model output without any of that, so they can't tell you what users are shown.
Do I get the SSE stream and an HTML file?
You always get the SSE stream, which carries the complete response. HTML appears only when an engine returns HTML instead of SSE, and the engine decides that per request.
What structured modules do you capture?
Answer text, citations, and fan-out / follow-up queries, plus ads on ChatGPT. Shopping carousels and "People Also Ask" blocks vary by engine. Tell us the elements you track and we'll confirm coverage per engine.
Tech & Reliability
Async or sync?
Both, at no extra cost. For async you POST /v1/wire/task and get a job_id back immediately, then poll GET /v1/wire/jobs/{id} for the clean result once it's ready. One task per submission.
What concurrency and volume can you handle?
We offer the best concurrency in the market, up to 600 concurrent and up to 1,500 RPM, and we can raise it further for larger contracts. Bursts and scheduled batches are fine.
What's your success rate and latency?
Engines run at ~99.8% success with p95 under 30s, at ~1M requests/day for clients.
Do you support webhooks?
Yes. Async jobs support HMAC-signed webhooks, so results are pushed to you when they're ready. Polling the job endpoint works too.
How do you handle redundancy and outages?
Every engine runs across a large proxy pool with automatic retries, so one blocked route doesn't fail your job. Breakages are picked up by our engineering team, and failed jobs are never billed.
Do you offer an SLA?
Yes, on enterprise contracts, with uptime and support terms agreed up front. Talk to us for details.
What happens on a failed request?
You're not charged. Only successful requests are billed. We retry internally before marking a job failed, and failures never use credits.
What if an engine changes its UI and breaks collection?
We self-heal. Automated checks detect UI changes and adapt, backed by a 25+ person engineering team that fixes breakages fast, so your feed doesn't go quiet. Contracts of $1k+/month get priority support.
Can I control the country the prompt runs from?
Yes, set the country per request and we route through a proxy there, so the response reflects that location. State and city are supported in major markets, at no surcharge. The country list is largely shared across engines, with a few small markets marginally narrower on some engines.
Are responses unbiased? Do memories leak between requests?
No. Each prompt runs in a fresh conversation(optionally incognito), and context is never shared between calls or customers.
Pricing & Credits
Pay-as-you-go or subscription?
Monthly plans, from $100/mo, that you can change or cancel anytime. Start on the 300 free credits, pick the volume you need, and your per-1K rate drops as you scale. Larger volumes can move to custom terms.
Any hidden fees or add-ons?
None. Geo-targeting, async and sync endpoints, and web search with citations are all included at no extra cost.
Is pricing the same across engines?
Standard answers cost the same on every supported engine (ChatGPT, Gemini, Perplexity, AI Mode, AI Overviews, Naver, Wenxin). ChatGPT ads data is priced separately.
Is there a minimum commitment?
No long-term contract. Paid plans are monthly from $100, and you can change or cancel anytime.
How do credits work?
You get 300 free credits on signup, about 150 answers at 2 credits each. Paid plans are counted in answers, where one answer is one engine response. Free-credit cost per request:
Getting Started
How do I test with no commitment?
Sign up with Google and get 300 free credits to run every engine right away. Need more test volume? Any monthly plan works, and you can cancel anytime.
How do I get an API key?
From your dashboard, go to Settings → API Keys and generate one. Pass it as the X-API-Key header on every request.
Security & Legal
Are you SOC 2 compliant?
Yes. Anakin is SOC 2 Type II and ISO 27001:2022 certified, and GDPR compliant.
Where do your proxies come from?
Requests route through our proxy network across 207 countries and territories, matched to the country you set. For vendor or security reviews, we'll walk you through sourcing and our controls.
Do you certify compliance with the engines' Terms of Use?
We operate as the data-collection infrastructure layer and don't issue a formal ToS-compliance statement. The terms-of-use assessment is left to each customer's own use case.
Do you provide legal indemnity?
No formal indemnity is included. We're happy to walk through how customers typically approach this.
How are you better than BrightData / Firecrawl / DataForSEO / Exa / Cloro?
One clean API instead of stitching several vendors, and we cover the AI answer engines with real answers + citations, which SERP and page scrapers don't. ~1M requests/day at 99.8%, top-tier concurrency, self-healing on UI changes. Typically 30–50% cheaper, and often cheaper on credit consumption too, you only pay for successes.
Build on real AI answers.
Answers, citations and ChatGPT ads from every major engine, in one API call. Start free, scale when it works.