Skip to content
FuelTurin · Wave by Vento side event · Turin

The Fuel for
Web Scraping & AI Agents

Reliable and secure proxies for seamless access, anonymity, and performance worldwide.

Test our proxies with a free trial. No credit card required.
Trusted by
Proxy infrastructure

Proxies we own and operate

Residential, mobile, ISP and datacenter IPs. Consent-sourced, GDPR-compliant, with city and ASN targeting. Pick by trust level and budget.

195+countries & cities
MillionsIPs, refreshed daily
99.9%uptime
SOC 2 Type IISOC 2 Type IIIn progressISO/IEC 27001ISO/IEC 27001In progress
Enterprise security controls, independently audited.
Web data infrastructure

APIs that return the data, not the page

Extract, crawl, map, search and LLM-structured output from any site. Unlocking, rendering and CAPTCHAs handled inside the request. Pay per success.

5APIs, one key
$0for failed requests
Millionspages per crawl
SOC 2 Type IISOC 2 Type IIIn progressISO/IEC 27001ISO/IEC 27001In progress
The same audited controls behind every API request.

Why Datafuel

Straight from the source

Most providers resell bandwidth they buy wholesale. We built the network, the SDK and the APIs. That is why the IPs are clean and the price is honest.

From IP to structured data

Proxies, scraping APIs and web agents on one platform. One account, one balance, one invoice. Most providers sell these separately and let you integrate them.

24/7 Support

Live Chat, Discord, Telegram, Slack. You reach people who can actually fix it, backed by 99.9% uptime, so you rarely need to.

Built in Europe

GDPR and the AI Act are our home ground, not a compliance exercise. You can document where every IP comes from, because we can.

Integrations

Works with your stack

Official SDKs, native MCP support, and a live playground for every product.

SDKs & API

Ten lines, any language

Pick a product and a language. Click a highlighted value in the request to change it — the response updates live.

import requests
user = "-".join([
"network-residential",
"country-us",
"region-california",
"city-losangeles",
"os-windows",
"session-394893",
])
proxy = f"http://{user}:gelato492@gw-us.datafuel.ai:5555"
r = requests.get(
"https://ipinfo.io/json",
proxies={"http": proxy, "https": proxy},
)
print(r.json())
{
"ip": "172.58.14.203",
"city": "Los Angeles",
"region": "California",
"country": "US",
"org": "AS7922 Comcast"
}
AI & Agents

Give your agent the whole web

One config. Connect your agent to Datafuel via MCP. Unlocker, LLM Scraper, Crawl and Map become tools it can call.

{
"mcpServers": {
"datafuel": {
"command": "npx",
"args": ["-y", "@datafuel/mcp"],
"env": {
"DATAFUEL_API_KEY": "df_key_your_key_here"
}
}
}
}
Works withMCMCPCLClaudeCUCursorLALangChainand any MCP client
View the docs

Agent onboarding. Are you an AI agent? Fetch this file for the whole API surface, get a key and start building.

curl -s https://datafuel.ai/llms.txt
Official SDKsNONode.jsGOGoPYPython · soon
Read llms.txt
FAQ

Frequently asked questions

Straight answers, including the uncomfortable ones.

Proxy infrastructureWhere the IPs come from and what happens under pressure.
Where do your residential IPs come from?
From real devices whose owners opted in through the Datafuel SDK, embedded in partner apps that disclose the bandwidth sharing and reward the user for it. Every IP is consent-based, can be withdrawn by the user at any time, and the network runs under GDPR. You can document the provenance of every request, because we can.
What happens when a target blocks me?
Nothing you have to handle by hand. Proxy rotation retries on a fresh IP automatically, the Unlocker API solves CAPTCHAs inside the request, and you only pay if the retry succeeds. The proxy tier and JavaScript rendering are your choice per request; Datafuel never upgrades them on its own. Persistent blocks appear in your dashboard with the target and the reason.
Which countries and locations do you cover?
Residential IPs across 223 countries and territories, mobile in 194, with country, city and ASN targeting; ISP and datacenter IPs in the main EU and US hubs. The full list lives on the Locations page. If you need a specific city or carrier, ask and we will tell you the real availability, not a marketing number.
Scraping APIsWhat comes back, and how it behaves on hard targets.
Which formats do the scraping APIs return?
Markdown, structured JSON (from a schema or a plain-language prompt), raw HTML and screenshots. The Unlocker returns one page, Crawl and Map return URL lists with per-page results, and Search returns structured SERP data. Pick the format per request.
Do the APIs handle JavaScript-heavy sites and CAPTCHAs?
Yes. Every request runs through our own proxy network with automatic retries and solves CAPTCHAs on protected targets. For JavaScript-heavy pages set js_rendering: true; if a page needs it and you did not ask, the API says so when the task is created, so you never pay for a render you did not request. You only pay when the response is usable.
Can my AI agent use the APIs directly?
Yes. The MCP server gives Claude, Cursor, LangChain and any MCP client tools for the Unlocker, LLM Scraper, Map and Crawl, plus jobs, balance and usage, and llms.txt gives an agent the whole API surface in one file. No SDK required.
Legal & complianceWhat you may collect, and what you may bring of your own.
Is scraping legal? What are we allowed to collect?
Collecting publicly available data is lawful in the EU and the US in most cases; what matters is how you collect it and what you do with it. You may not bypass logins, collect personal data without a lawful basis, or breach terms you have agreed to. Our Acceptable Use Policy spells this out, and we review enterprise use cases during onboarding.
Can I use my own IPs, or bring my own AI model key?
Yes on both. AI parsing on the Unlocker API uses your own OpenAI, Anthropic or Google key, so extraction runs on your model contract and your data terms. Routing IPs you already own through our gateway is available on enterprise plans; talk to us and we will scope it.
PricingWhat you pay for, and what you never pay for.
How is pricing calculated? What counts as a successful request?
Residential and mobile proxies are billed per GB of traffic, ISP and datacenter proxies per IP. Scraping APIs are billed per successful request, and a request counts as successful only when we return a 2xx response with usable content. Blocks, CAPTCHAs, timeouts and errors are free. No seat fees and no minimums on self-serve plans.
Do you offer a free trial?
Yes. The scraping APIs come with 1,000 free credits, once per account, valid for 1 year, no card required. Proxy networks have a free trial on request through live chat. Failed requests are never charged, and unused plans can be refunded within 24 hours of purchase.
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.