Locus product guide · Updated November 10, 2026

URL to Markdown API: Jina Reader, Firecrawl and others

A URL to Markdown API fetches a web page and returns its main content as clean Markdown for an LLM. Jina Reader works by putting r.jina.ai/ in front of any URL and is free at 20 requests a minute without a key. Firecrawl charges 1 credit per page, with 1,000 free a month. Parallel Extract bills per URL, and Cloudflare's /markdown endpoint bills browser time at $0.09 an hour after 10 included hours on Workers Paid. Prices checked November 10, 2026.

What does a URL to Markdown API do?

It loads the page, runs its JavaScript when needed, strips menus, ads and scripts, and returns the readable content as Markdown. An agent can drop that text straight into its context.

The services differ in three ways: whether they render JavaScript, how they bill (per page, per token or per browser minute), and whether they can return structured JSON instead of Markdown.

How do the main URL to Markdown APIs compare?

Figures come from each vendor’s pricing page or docs. Prices checked November 10, 2026.

How do the main URL to Markdown APIs compare?
Service	Free allowance	How you pay	Catch
Jina Reader	20 requests a minute with no key. 10 million free tokens on each new key.	Prepaid tokens, billed on the tokens in the output	Long pages cost more. Average latency is listed at 7.9 seconds.
Firecrawl scrape	1,000 credits a month, 10 scrapes a minute	1 Firecrawl credit per page, on monthly plans by credit volume	JSON output adds 4 credits per page. Credits roll over only on Scale.
Cloudflare Browser Run /markdown	10 minutes of browser time a day on Workers Free	Workers Paid ($5 a month minimum) includes 10 hours, then $0.09 per hour	Billed by browser time, not pages. JavaScript-heavy pages may need wait options.
Parallel Extract	$5 of usage a month with a card on file	Per URL, billed on usage	Returns excerpts focused on an objective by default. Full content is optional.
Tavily Extract	1,000 credits a month	Basic depth: 1 Tavily credit per 5 URLs, pay as you go	Advanced depth doubles the cost. Failed URLs are free.
Diffbot Extract	10,000 credits a month, 5 requests a minute	1 Diffbot credit per page, on monthly plans	Returns article fields as JSON rather than Markdown

How is URL to Markdown priced through Locus Pro?

Per call, from one prepaid balance, with no Firecrawl, Parallel or Diffbot account. Locus Web Extract is priced per URL, and most other extract endpoints by request. The endpoints below come from the Locus catalog on November 10, 2026.

Locus Web Extract reads up to 20 URLs per call and retries any failed URL with a second extractor. Jina Reader is not in the catalog. Some unreviewed third-party listings, such as Skim read, also return Markdown.

How is URL to Markdown priced through Locus Pro?
Locus endpoint	How it is priced	Returns
Locus Web Extract	Per URL, with a ceiling set by maxCredits	Markdown or JSON per URL
Firecrawl scrape v2	By request	Markdown by default
Diffbot article	Fixed price per call	Article text and fields as JSON
Parallel extract	By request, up to 20 URLs	Markdown excerpts or full page
Tavily extract	By request size	Parsed text

How far do the free tiers go?

For light use, a free tier may be all you need. Parallel gives $5 of usage a month with a card on file. Firecrawl’s free plan covers 1,000 pages a month, and Diffbot’s 10,000 credits a month. Jina Reader needs no account at all.

Free tiers also cap speed. Jina allows 20 requests a minute without a key, Firecrawl’s free plan 10 scrapes a minute and Diffbot’s 5 requests a minute, so a large batch outgrows them quickly.

What makes URL to Markdown conversion fail?

Check these before you trust the output in an agent loop.

JavaScript: single-page apps can return empty text unless the service waits for rendering. Cloudflare documents load options, such as gotoOptions, for this case.
Logins and paywalls: none of these services sees content behind your login unless you pass cookies or credentials, which Cloudflare supports.
PDFs: Firecrawl charges 1 credit per PDF page, so a 40-page report costs 40 Firecrawl credits.
Token billing: Jina bills on output tokens, so one long page can cost as much as many short ones.
Site terms: some sites forbid automated access. The API does not change that.

How do I turn URLs into Markdown from an agent without code?

Connect the agent to Locus Pro and give it the links. Tell it: "Install Locus Pro by following https://paywithlocus.com/SKILL.md". Then ask: "Read these five URLs and give me each page as Markdown."

The agent picks an extract endpoint, checks the price for free, and runs it with a credit ceiling. Locus Web Extract never searches or swaps in other pages, so you get exactly the URLs you asked for.

Which URL to Markdown API should I use?

Match the tool to your volume and your pages.

Occasional use from a script: Jina Reader with no key, or Firecrawl’s free plan.
Steady volume in production: a Firecrawl plan or Parallel Extract.
Frequent Tavily Extract calls: Tavily itself, since it can cost less per call with your own Tavily account and API key.
Your stack already runs on Cloudflare Workers: the /markdown endpoint, billed by browser time.
News and blog articles with author and date fields: Diffbot.
An agent that reads a few pages between searches and other calls: Locus Web Extract, with a credit ceiling on each call.

Frequently asked questions

Is Jina Reader free?

It works without a key at 20 requests a minute. A free key raises the Reader limit to 500 requests a minute and comes with 10 million tokens. After that you buy tokens.

Can I send raw HTML instead of a URL?

Yes, with Cloudflare’s /markdown endpoint, which accepts either a url or an html field. Diffbot and Firecrawl are built around URLs.

Do failed pages cost money?

It depends on the vendor. Tavily never charges for a failed URL extraction. Cloudflare does not bill a Quick Action that fails with a waitForTimeout error. Check each vendor’s rule before a large batch.

Which option returns structured JSON as well as Markdown?

Firecrawl adds a JSON format for 4 more credits per page. Locus Web Extract returns JSON per URL when you pass a prompt or schema. Diffbot always returns JSON fields.

Implementation references

Use these first-party references for current request contracts and account requirements. Tool availability, prices, and negotiated terms can change.

Sources

Third-party details were checked on November 10, 2026. Vendors change pricing and features often, so confirm on their site before you decide.