Orpheus — Requests that get through

The problem

Your scraper works.
Until it doesn't.

One morning the site you depend on adds a bot wall. The data stops. Someone on your team spends the next two weeks reverse-engineering it — and the vendor ships a new version the week after.

Headless browsers eat your budget

Hundreds of MB of RAM per page, seconds per request, and they still get flagged.

Every protection is a different puzzle

Sensors, sliders, proof-of-work, tokens bound to your IP. Each one is its own rabbit hole.

It breaks silently

A cookie expires, a fingerprint drifts, an IP gets burned. You find out when the dashboard goes empty.

Your engineers hate it

Nobody joined your team to babysit anti-bot updates.

How it works

You ask for a page.
You get the page.

Send your request

Exactly like any HTTP client. Same URL, same headers, same code you'd write without a wall in the way.

We clear the wall

The challenge is recognised and solved behind the scenes, with an identity the site trusts end to end.

You get the response

The original request is replayed and its answer handed back. The session stays valid for the next ones.

Why teams pick it

Built for people who
need the data on Monday.

~5 s

to clear a Cloudflare wall

Measured on real sites: 14 out of 15 cleared, about five seconds each. Then plain HTTP speed for every request after that.

retry loops for you to write

Detection, solving, cookie handling and replay are one call. If a route fails, the next one takes over on its own.

One endpoint

Works from any language — curl, Python, Node, Go. Send JSON, get the page back.

Looks like a real browser

Every request carries a consistent, genuine browser identity — not a script pretending.

Tuned per site

Routing is chosen from real-world benchmarks, and you can override it site by site.

The API

Send a URL. Get the page.

Nothing to install or host. Your proxy, our solving — one call from any language.

CLOUD API · INVITE ONLY

Orpheus API

For teams who just want the page, without the anti-bot rabbit hole.

One endpoint, any language
Challenge solving included — just bring your proxy
Sticky sessions for multi-step flows, with your own starting cookies
Dashboard with usage, API keys and request logs

Reserve my API key Read the docs

curl https://orpheus-api.lapetiteagora.fr/v1/fetch \
  -H "Authorization: Bearer $KEY" \
  -d '{"url": "https://shop.example/p/42",
       "proxy": "http://user:pass@host:port"}'

# → 200
{
  "status": 200,
  "body": "<!doctype html>…",
  "solved": ["datadome"],
  "session_id": "s_Ij4oudr1…"
}

	Headless browser	DIY reverse engineering	Orpheus
Time to first page	Hours	Weeks	Minutes
Cost per request	High (CPU + RAM)	Low, once it works	Low
Speed after the wall	Seconds per page	HTTP speed	HTTP speed
When the wall changes	Often flagged anyway	Back to square one	Handled for you
Maintenance	Medium	Constant	None on your side

FAQ

Questions, answered.

Do I need proxies?

Yes — bring sticky residential or ISP proxies. A session keeps the same proxy from the first challenge to the last request, which is what keeps the clearance valid.

Which protections are covered?

Akamai, DataDome, Imperva, Kasada and Cloudflare challenges, plus reCAPTCHA, Turnstile and AWS WAF tokens. Coverage grows with every release.

Which languages can I use?

Any language that can send an HTTPS request. You post a small JSON object and get the page back as JSON — examples in curl, Python and Node are in the documentation.

How much will it cost?

Pricing is being finalised with early users. People on the waitlist get founding-member pricing and a free allowance to try it on their own targets.

Can I use it on any website?

You're responsible for respecting the terms of the sites you access and the laws that apply to you. We'll ask about your use case when onboarding.

Early access

Get through
before everyone else.

We're onboarding a small group of teams first. Join the list for early access, founding-member pricing and a direct line to the people building it.

You're on the list.

We'll reach out personally when your access is ready.

Your scraper works.Until it doesn't.