URLpipe vs Jina Reader
URLpipe vs Jina Reader
Both turn a URL into clean, LLM-ready Markdown and both serve agents over MCP. Jina Reader is a URL prefix billed by the token; URLpipe is nine endpoints, including /scrape, at a fixed price per call. Here's how they compare — and when Jina is the better pick.
By Roger Campos · Last updated: September 2026
TL;DR
Choose Jina Reader for a no-signup URL prefix, PDF reading, image captions and per-token pricing on short pages. Choose URLpipe when you want a fixed price per call, Markdown made without a model, free cache hits for up to 30 days, and screenshots, metadata, Lighthouse audits and console errors from the same key — or several from one page visit.
Free plan, no credit card. 1,000 credits a month.
At a glance
URLpipe vs Jina Reader, feature by feature
Both render JavaScript, take CSS selectors and return clean Markdown. Jina leads on friction and documents — the prefix, PDFs, image captions. URLpipe leads on breadth per URL and on what a call costs.
| Capability | URLpipe | Jina Reader |
|---|---|---|
| Clean Markdown output | Yes | Yes |
| Renders JavaScript (real browser) | Yes | Yes |
| How Markdown is made | Deterministic walk of the rendered DOM | Readability-style extraction |
| Try it with no signup | Partial | Yes |
| No API key for basic use | Partial | Yes |
| CSS-selector targeting / exclusion | Yes | Yes |
| Wait for an element before reading | Yes | Yes |
| Block ads, remove cookie banners | Yes | Not documented |
| Full-page screenshots | Yes | Yes |
| Screenshot options (element, dark mode, WebP, hidden selectors) | Yes | Not documented |
| Metadata extraction endpoint | Yes | Partial |
| AI summary endpoint | Yes | No |
| Keyword extraction endpoint | Yes | No |
| Google Lighthouse audit | Yes | No |
| JavaScript console / error capture | Yes | No |
| Several outputs from one page visit | Yes | No |
| PDF input | No | Yes |
| Image captions (vision model) | No | Yes |
| Hosted MCP server | 14 tools | 12 tools |
| Async + signed webhook | Yes | No |
| Idempotency key, labels, request history | Yes | No |
| Cache | Up to 30 days; hits are free | About 5 minutes |
| Residential-IP fetching | One exit, per request | Proxy by country, or your own |
| EU processing, AI included | Yes | Experimental EU endpoint |
| Self-hosting | No | Apache-2.0 repo; ReaderLM-v2 is CC-BY-NC |
| Billing unit | Credits, fixed per endpoint | Tokens in the response |
URLpipe's “partial” on signup: the free tools need no account, the API needs a key. Jina's “partial” on metadata: its frontmatter modes return basic fields, with no dedicated endpoint. “Not documented” means we found nothing in Jina's docs. Checked against jina.ai/reader and Jina's GitHub on September 24, 2026.
What changed around Jina
Jina Reader is part of Elastic
Elastic acquired Jina AI in October 2025. Reader carries on as before at r.jina.ai, with a hosted MCP server at mcp.jina.ai and an experimental EU endpoint at eu.r.jina.ai. What moved is licensing: commercial and on-premises use now goes through Elastic's sales team, and the ReaderLM-v2 model is licensed CC-BY-NC-4.0, which rules out commercial use without that licence. The jina-ai/reader repository itself is Apache-2.0.
None of that makes Reader worse at its job. It does change the question some teams ask — “who will we be buying from in a year, and on what terms?” — which is often when people start looking at alternatives.
How they differ
Four differences that actually matter
Friction: a URL prefix vs an authenticated POST
Jina's signature move is the prefix: put r.jina.ai/ in front of a URL and read the Markdown in your browser, no account needed at 20 requests a minute. URLpipe's free tools give you the same in a browser with no account; the API takes a free sign-up and a bearer token on a POST. The key buys per-project request history, labels, idempotency keys and async delivery.
Markdown: made the same way every time
URLpipe's /markdown doesn't call a model. It walks the rendered DOM and writes down the main content — headings, lists, links, tables, code — in about 20 ms, identically for the same page. On a 33-page corpus it kept 87.2% of the visible text, and 11.0% of its output was text a visitor never sees. The Readability strategy Jina's Reader is built on, run on the same pages, scored 88.2% and 21.3%.
Billing: tokens vs a fixed price, and a cache that lasts
Jina bills the tokens in its response, so a long page costs more than a short one, and its cache holds results for about five minutes. URLpipe charges 1 credit for Markdown whatever the page's length; repeat requests within max_age (7 days by default, up to 30) are free, and an identical request arriving while the first is still running waits for it for free.
Breadth: text vs everything about the page
Jina does text very well, plus PDFs, image captions and a viewport or full-page screenshot. URLpipe turns the same URL into Markdown, rendered HTML, a full-page screenshot with options, metadata, a summary, keywords, a Lighthouse audit or its console errors — and /scrape takes several from one page visit. If PDF ingestion or captioning is central, Jina has the edge.
The benchmark reimplements each strategy on the same fetched DOM; it measures the approaches, not the vendors' live APIs.
Switching
What moving a Jina Reader call looks like
A Jina call is a GET with the target in the path and options in headers. The URLpipe equivalent is a POST with the target and options in a JSON body. Add "sync": true to get the Markdown back in the response, as Jina does; leave it out and you get a token straight away, with the result at your webhook or at GET /result/:token.
curl https://r.jina.ai/https://example.com/pricing \
-H "Authorization: Bearer $JINA_API_KEY" \
-H "X-Remove-Selector: .cookie-banner, #chat"curl -X POST https://urlpipe.dev/markdown \
-H "Authorization: Bearer $URLPIPE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"url": "https://example.com/pricing",
"sync": true,
"page_options": {
"block_cookie_banners": true,
"remove_selectors": ["#chat"]
}
}'X-Remove-Selector→page_options.remove_selectors;X-Wait-For-Selector→page_options.wait_for_selector.X-Respond-With: html→ /html;screenshotorpageshot→ /screenshot, which is full page by default.X-No-Cache→"max_age": 0. Anything younger thanmax_ageis served free.- PDFs and
X-With-Generated-Althave no URLpipe equivalent — keep Jina for those.
Pricing
Free tiers compared
Both are usable for free. They meter differently — Jina by the tokens it returns, URLpipe by fixed credits per endpoint with free cache hits. Jina's figures are from jina.ai/reader, September 24, 2026.
URLpipe — Free
$0 / month
- 1,000 credits a month — 1,000 Markdown conversions, or 1,000 screenshots, or any mix
- No credit card; the free tools need no account at all
- Cache hits are free for up to 30 days
- Paid plans from $19/mo, never cut off: past the allowance it's $1.50 per 1,000 credits
Jina Reader — Free
$0 to start
- 20 requests/minute with no API key
- 10 million free tokens for every new key, shared across Jina's APIs
- 500 requests/minute with a key
- Then billed by tokens in the response, in prepaid packages
- Cache holds results about 5 minutes
Who it's for
Pick the right tool for the job
Use URLpipe if you…
- Want to know what a call costs before you make it, whatever the page's length
- Need screenshots, metadata, summaries, keywords, Lighthouse or console errors next to the Markdown
- Re-read the same URLs and want cache hits to cost nothing
- Want async delivery, idempotency keys and requests labelled by customer
- Need AI processing kept in the EU, from a hosted service
Use Jina Reader if you…
- Want the no-signup URL prefix, straight from a browser or a shell
- Read PDFs, or want images captioned by a vision model
- Mostly fetch short pages, where per-token pricing is cheap
- Want to self-host the Apache-2.0 Reader repository
- Already use Jina's search, embeddings or reranker
The verdict
Is URLpipe a good Jina Reader alternative?
Yes — when Markdown alone isn't the whole job, when you'd rather budget in calls than tokens, or when you want Markdown produced the same way every time without a model. Jina Reader stays excellent for the frictionless prefix, for PDFs and for image captions. If you only feed clean text to a model, either works; once you need more than text about each page, URLpipe puts it behind one key.
FAQ
Frequently asked questions
Is URLpipe a good Jina Reader alternative?
What is the difference between URLpipe and Jina Reader?
Who owns Jina Reader?
Can I self-host Jina Reader commercially?
How does pricing compare?
Does URLpipe support CSS selectors like Jina's X-Target-Selector?
Can I try URLpipe without signing up?
More than Markdown, from one API.
Markdown, screenshots, metadata, summaries, keywords, Lighthouse and console errors — over HTTP or MCP. 1,000 credits a month, no card.