Unlocker API
One call returns any page past any protection, as Markdown or structured JSON. Anti-bot handling, proxy rotation, CAPTCHA solving and JavaScript rendering all happen inside the request.
- Cloudflare, Akamai, DataDome, PerimeterX and every major CAPTCHA, handled inside the call
- 99.6% success on protected targets
- Proxies are automatically managed: tier, rotation and retry-on-ban
- JavaScript rendering with waits and page actions
- HTML, Markdown, screenshot or JSON output, auto-parsed fields
- AI parsing with your own model key
Sign up with GoogleMCP readyConnect your agent in one command:
npx -y @datafuel/mcp initwww.amazon.com
United States/s?k=wireless+headphones
1Request
2Retrieve
3Format
4Ready
www.amazon.com
Opening the source page
One request starts the example.
Everything in one request
The scannable summary. Deep dives below.Anti-bot & WAF bypassCloudflare, Akamai, DataDome, PerimeterX, Imperva. Fingerprints and TLS handled per request.
CAPTCHA solvingreCAPTCHA, hCaptcha, Turnstile, DataDome slide: solved inside the call, no third-party solver.
Rotation & retry-on-banA blocked IP is swapped and the request retried before it reaches your code.
Basic / Premium / Mobile tierPick the proxy tier per request; escalate automatically when a cheaper exit is blocked.
Geo-targetingproxy_country routes the request through 195+ countries; state and city on Premium.
Sticky sessionsproxy_session_id keeps the same IP across several requests for logins and carts.
JavaScript renderingA real browser with wait_for_selector, wait_for and page actions before extraction.
Headers, cookies, deviceCustom headers, cookies, user agent and device profile on every call.
Seven output formatsHTML, Markdown, text, JSON, screenshot, PDF or structured fields from one result_format.
AI parsing, your keyA prompt and a JSON schema turn any page into fields; tokens billed on your provider key.
Async bulk jobsPOST /job with hundreds of URLs, multithreaded, with an id at once and paged results.
Pay per success, analyticsFailed requests are free; usage per key and per endpoint in the dashboard.
DatafuelUnlocker API · AI parsing
Processing
JSON ready
AI providerOpenAI
AnthropicGeminiMistral
OpenAIAnthropicGeminiMistral›Extract name, price, currency, rating and availability as JSON
Opening the shopping result
Extracting the requested fields
Sending to your model
shopping.google.com · product result
Loading
HTTP 200
nameAir Jordan 1 Retro High OG, Size 43price · currency€189.99rating★★★★★4.8 (6,120)in_stockIn stocksneakershop.example · free delivery · demo data
Sent to OpenAI on your keyai_provider: openai
Your JSON is ready
{
"name": "Air Jordan 1 Retro High OG, Size 43",
"price": 189.99,
"currency": "EUR",
"rating": 4.8,
"in_stock": true
}
Datafuel opens the result page through a premium residential exit.
The fields named in your prompt are located on the page.
Prompt and page go to your provider, on your key. The model shapes the answer.
The same values, as the JSON you asked for.
01 · AI parsing
AI parsing with your own API key
Send a prompt and a JSON schema with the request. The page is fetched through our network, then handed to your model on your key: OpenAI, Anthropic, Google or Mistral. You get the fields, not the HTML, and the model tokens are billed by your provider, not by us.
1curl -X POST https://scraping-api.datafuel.ai/api/v1/task \2 -H "X-API-Key: $DATAFUEL_API_KEY" \3 -H "Content-Type: application/json" \4 -d '{5 "type": "unlocker",6 "proxy_type": "Basic",7 "proxy_country": "US",8 "attributes": {9 "url": "https://example.com/product/123",10 "method": "GET",11 "result_format": "json",12 "result_use_ai": true,13 "result_ai_prompt": "Extract product name, price and stock status",14 "result_ai_format": {15 "title": "string",16 "price": "number",17 "in_stock": "boolean"18 },19 "ai_provider": "openai",20 "ai_model": "gpt-4o-mini",21 "ai_api_key": "$OPENAI_API_KEY"22 }23}'
DatafuelUnlocker API · anti-bot & CAPTCHA
Processing
200 OK
urltripadvisor.com/Hotel_Review-demoproxy_typeautosessionrotate
Demo data · illustrative scenario. The five challenge types are shown in sequence; a real request meets whichever one the target serves.Challenge 1 of 5 · Cloudflare Turnstile
Challenge 2 of 5 · Cloudflare Challenge
Challenge 3 of 5 · AWS WAF
Challenge 4 of 5 · reCAPTCHA
Challenge 5 of 5 · Microsoft
tripadvisor.com/Hotel_Review-demo
Challenge page
Verify you are human
Success!
challenge page · solving inside the request
cleared · the original request continues
Checking your browser before accessing the site…
Verification passed
challenge page · solving inside the request
cleared · the original request continues
Verifying that you are not a robot…
Verification complete
challenge page · solving inside the request
cleared · the original request continues
I’m not a robot
Verified
challenge page · solving inside the request
cleared · the original request continues
Prove you’re human
Verified
challenge page · solving inside the request
cleared · the original request continues
Cloudflare TurnstileCloudflare ChallengeAWS WAFreCAPTCHAMicrosoft
Listing delivered · 200 OK
{
"status": 200,
"url": "https://tripadvisor.com/Hotel_Review-demo",
"challenge": "cloudflare_turnstile · cleared",
"fingerprint": "ja3 · fresh",
"proxy": "basic → premium · 92.118.4.201",
"attempts": 2,
"billed_requests": 1,
"markdown": "# Grand Harbour Hotel\n4.5 · 2,318 reviews · from €142 / night…"
}
Cloudflare Turnstile is solved inside the request, with a TLS fingerprint that is new on every call. Your code never sees the widget.
Cloudflare’s managed challenge (the “checking your browser” page) passes the same way, on a rotating IP.
AWS WAF challenge pages are verified and the original request continues.
reCAPTCHA, every version: solved automatically. No third-party solver to wire up.
Microsoft’s human check, cleared. If a Basic exit is blocked, the retry goes out on Premium; pin one IP with proxy_session_id.
200 OK, one call billed. This request met Turnstile; the same call clears the other four. Failed attempts are free.
02 · Protections
Anti-bot and proxy management
Cloudflare, Akamai, DataDome, PerimeterX and every major CAPTCHA are cleared inside the request: fresh TLS fingerprint per call, rotation on ban, escalation from Basic to Premium when a datacenter exit is blocked. proxy_session_id pins one IP across calls when a login or a cart needs it.
1curl -X POST https://scraping-api.datafuel.ai/api/v1/task \2 -H "X-API-Key: $DATAFUEL_API_KEY" \3 -H "Content-Type: application/json" \4 -d '{5 "type": "unlocker",6 "proxy_type": "Premium",7 "proxy_country": "GB",8 "proxy_session_id": "checkout-8f3a",9 "attributes": {10 "url": "https://example.com/account/orders",11 "method": "GET",12 "result_format": "json"13 }14}'
DatafuelUnlocker API · bulk jobs
Processing
Results ready
POST/api/v1/joburls512 URLs · amazon · walmart · tripadvisor · google …multithreadedtrue · concurrency 75proxy_typeBasic
Endpoints and fields follow the public OpenAPI spec; the pace is not measured latency.POST /job · 512 URLs accepted, id returned at once
Polling GET /job/{id} · 75 tasks in flight
Completed · 512 of 512 tasks
job9f2c7a1e-4b3d-4e7a-9c21-demo
created
processing
completed
tasks_count
512
512
512
512
512
512
tasks_done
0
96
211
318
437
512
running
0/ 75
75/ 75
75/ 75
75/ 75
75/ 75
0/ 75
credits_used
0
96
211
318
437
512
progress512 tasks
concurrency75 of 100 allowed
queued
processing
done · 1 credit
queued
processing
done · 1 credit
queued
processing
done · 1 credit
queued
processing
done · 1 credit
queued
processing
done · 1 credit
queued
processing
done · 1 credit
… 506 more tasks
GET /job/{id}/results
{
"tasks_count": 512,
"tasks_complete": 512,
"tasks_failed": 0,
"credits_used": 512,
"tasks_result": [
{ "id": "c41e…", "result": "# Wireless Noise-Cancelling Headphones…" },
{ "id": "7b09…", "result": "# Pokémon 151 Elite Trainer Box · $89.97…" },
{ "id": "e2d5…", "result": "# Grand Harbour Hotel · 4.5 · 2,318 reviews…" },
… 509 more
]
}
One POST with 512 URLs across many sites. The API answers with a job id before anything is scraped.
Poll GET /job/{id}. With multithreaded: true, 75 tasks run at once, under your plan's limit of 100; counters move as each one finishes.
status: completed. 512 tasks, nothing left to wait for.
GET /job/{id}/results returns one entry per URL. Credits are charged per completed task; failed tasks are not.
03 · Bulk jobs
Bulk jobs, asynchronous
Hundreds of URLs across many sites in one POST /job. The id comes back at once, tasks run in parallel up to your plan's concurrency, and GET /job/{id}/results returns one entry per URL. An Idempotency-Key makes retries safe; failed tasks are not billed.
1# 1. queue the batch → {"id": "9f2c7a1e-…"}2curl -X POST https://scraping-api.datafuel.ai/api/v1/job \3 -H "X-API-Key: $DATAFUEL_API_KEY" \4 -H "Idempotency-Key: batch-2026-09-13-a" \5 -H "Content-Type: application/json" \6 -d '{7 "type": "unlocker",8 "multithreaded": true,9 "proxy_type": "Basic",10 "attributes": {11 "urls": [12 "https://amazon.com/dp/B0DEMO123",13 "https://walmart.com/ip/demo-456",14 "https://tripadvisor.com/Hotel_Review-demo"15 ],16 "result_format": "markdown"17 }18}'1920# 2. poll progress → {"status": "processing", "tasks_count": 3, "tasks_done": 1, "tasks_remaining": 2, "total_cost": 1}21curl https://scraping-api.datafuel.ai/api/v1/job/$JOB_ID -H "X-API-Key: $DATAFUEL_API_KEY"2223# 3. fetch every result once status is completed24curl https://scraping-api.datafuel.ai/api/v1/job/$JOB_ID/results -H "X-API-Key: $DATAFUEL_API_KEY"
DatafuelUnlocker API · JavaScript rendering
Processing
Rendered
urlwalmart.com/browse/toys/pokemon-cards-demojs_renderingtruewait_for_selector[data-price]actionsclick "Load more"
Loading the page shell
JavaScript ran: prices are in
Action: click “Load more”
walmart.com/browse/toys/pokemon-cards-demo
JavaScript pending
Rendered
Pokémon TCG: Scarlet & Violet Booster Bundle
price loading…
$26.88
Pokémon 151 Elite Trainer Box
price loading…
$89.97
Charizard ex Premium Collection
price loading…
$49.88
Paldean Fates Premium Collection
price loading…
$49.99
Load more
Charizard ex Premium Collection$49.88
Paldean Fates Premium Collection$49.99
Rendered result · one request
{
"status": 200,
"rendered": true,
"wait_for_selector": "[data-price] · found",
"actions_run": ["click Load more"],
"products": 6,
"markdown": "# Pokémon cards\nScarlet & Violet Booster Bundle · $26.88 …"
}
A real browser loads the page. Images, fonts and media are never fetched.
The page's JavaScript runs; Datafuel waits until the prices exist before going on.
Your action runs before extraction, and the list grows in place.
The rendered content comes back as one request.
04 · Rendering
JavaScript rendering
js_rendering: true loads the page in a real browser, waits for the selector you name, runs your page actions (click, scroll, type) and only then extracts. Images, fonts and media are never fetched, so rendering stays fast and costs 5 credits instead of 1.
1curl -X POST https://scraping-api.datafuel.ai/api/v1/task \2 -H "X-API-Key: $DATAFUEL_API_KEY" \3 -H "Content-Type: application/json" \4 -d '{5 "type": "unlocker",6 "proxy_type": "Basic",7 "proxy_country": "US",8 "attributes": {9 "url": "https://app.example.com/product",10 "method": "GET",11 "js_rendering": true,12 "wait_for_selector": "#price",13 "result_format": "json"14 }15}'
Pricing
Pay only for successful requests
Pick a monthly plan, buy credits once, or run unlimited threads. Failed requests are never billed.
Every plan includes every endpoint and every interface (API, MCP). One shared credit balance: you only pay for volume, and only for successful requests.
How many requests a month?995,860
≈ 995,860 credits · Growth1K300K1M3.5M12MCustom
Free
Test every endpoint with real credits. No card, no expiry pressure.
$0/ mo, billed monthly
1,000 credits / month1,000 basic pages2 concurrent threads
- 1,000 trial credits
- Every endpoint, API + MCP
- Community support
GrowthBest fit
For production crawlers and agents that need headroom.
$79/ mo, billed monthly
1,000,000 credits / month1,000,000 basic pages25 concurrent threads$0.079 per 1K basic$0.395 per 1K JS
- Everything in Starter
- Auto top-up
- Priority chat support
Business
For teams shipping data products on a schedule.
$199/ mo, billed monthly
3,500,000 credits / month3,500,000 basic pages50 concurrent threads$0.057 per 1K basic$0.284 per 1K JS
- Everything in Growth
- Usage alerts per API key
- Account manager
Scale
High volume at the lowest per-page rate. Checks out instantly.
$499/ mo, billed monthly
12,000,000 credits / month12,000,000 basic pages150 concurrent threads$0.042 per 1K basic$0.208 per 1K JS
- Everything in Business
- Dedicated IP pool
- 99.9% SLA
Enterprise
Committed volume, invoicing and governance.
Custom
Custom credits / month250+ concurrent threads
- Custom volume pricing
- Custom concurrency (250+)
- SSO · audit log · DPA
Compare
| Free | Starter | Growth | Business | Scale | Enterprise | |
|---|---|---|---|---|---|---|
| Credits / month | 1K trial | 300K | 1M | 3.5M | 12M | Custom |
| Concurrent threads | 2 | 10 | 25 | 50 | 150 | 250+ |
| API keys | 1 | 3 | 10 | 25 | Unlimited | Unlimited |
| JS rendering | ||||||
| Residential proxies | ||||||
| AI in-flight (your key) | ||||||
| Geo-targeting · 195+ countries | ||||||
| Batch multi-URL | ||||||
| Unlocker · Crawl · Map · LLM Scraper | ||||||
| Auto top-up | — | — | ||||
| Support | Community | Priority chat | Account manager | Slack + phone | Dedicated + SLA |
What does a request actually cost?1 credit = 1 basic page · costs don't stack: JS + Residential is 20 credits, not 5 + 10
1×Basic HTTPSimple HTML, datacenter proxy
5×JS RenderingHeadless browser for dynamic sites
10×Residential proxyReal residential IPs
20×JS + ResidentialFull power for protected sites
FreeAI In-FlightLLM extraction on any Unlocker request, zero extra credits
FAQ
Frequently asked questions
The real objections, answered straight.
Requests & outputWhat you send and what comes back.
What counts as a successful request?
A response with the content you asked for: a 2xx from the target with a body, or, with AI extraction, a JSON object that matches your result_ai_format. Blocks, CAPTCHAs we could not solve, timeouts and empty bodies are failures and cost nothing.
Which output formats are supported?
json, csv, txt, markdown, png, jpeg and pdf via result_format. Markdown is the usual choice for LLM pipelines; set result_use_ai with a prompt and a JSON shape to get structured fields.
When do I need js_rendering?
Only when the page builds its content in the browser: infinite scroll, client-side prices, single-page apps. Try without it first, plain requests are 5x cheaper and faster. wait_for_selector tells the browser when the content is ready.
Do I need my own AI key?
No. AI extraction runs on Datafuel by default and is included in the request price. If you prefer your own provider and contract, pass ai_provider (openai, anthropic or google), ai_model and ai_api_key and the model runs on your account.
Protections & billingBlocks, limits and what you pay.
Which protections are supported?
Cloudflare, Akamai, DataDome, PerimeterX (HUMAN) and the major CAPTCHA types, plus generic anti-bot systems and rate limiters. They are handled inside the request; there is no flag to set per protection.
What happens when a target blocks me?
The Unlocker retries automatically on a fresh IP with a new fingerprint. If it still cannot get a usable response, the request fails and costs 0 credits. Per-domain analytics show success rate and cost for every target, so you see exactly where that happens.
Are there rate limits or a concurrency cap?
Concurrency is per account: GET /users/@me shows current_concurrency and concurrency_limit, and you can raise it from the dashboard or by talking to us. For large batches use POST /job, which queues URLs and can notify a webhook when the job completes.
Can I keep the same IP across several requests?
Yes. Pass a proxy_session_id and every request carrying that id exits from the same IP, which is what multi-step flows such as login, page and checkout need.
Get started
Ready to build?
Start with the free tier and scale as your project grows. No credit card, no sales call.
Talk to an engineer, not a chatbot.