Jina Reader
Prefix any URL with r.jina.ai and get clean, LLM-ready markdown back.
Overview
Jina Reader converts web pages, PDFs, and images into clean markdown for LLM pipelines: prepend https://r.jina.ai/ to any URL for content, or use s.jina.ai to fetch and convert the top web search results. The open-source branch of the service is Apache-2.0 and self-hostable via Docker.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 21 Jul 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
What you can produce with Jina Reader
- Turn any article, documentation page, or PDF into clean markdown by prepending https://r.jina.ai/ to its URL — no account needed at 20 requests a minute.
- Feed live web pages into a RAG pipeline or agent as token-efficient markdown instead of raw HTML.
- Ground an LLM with fresh web results via s.jina.ai, which searches and converts the top five hits in a single call.
- Extract readable text from PDFs, MS Office documents, and images (with AI captioning) through the same endpoint.
- Tune extraction with request headers — CSS selectors to target or exclude, timeouts, token budgets, and markdown or JSON response formats.
- Self-host the Apache-2.0 open-source build with Docker (ghcr.io/jina-ai/reader:oss, headless Chrome and LibreOffice included) when content must stay on your own infrastructure.
ASEAN Perspective
Jina Reader in Southeast Asia
ASEAN-region availability and pricing notes coming soon. Drop the editorial team a note via /contact/ if you can supply local context (Singapore/Malaysia/Indonesia/Thailand/Vietnam).
Jina Reader is the lowest-friction way to turn live web content into LLM-ready input. It converts URLs — web pages, PDFs, Office documents, images with AI captioning — into clean markdown via a simple URL prefix, and its search endpoint returns already-converted top results for retrieval grounding. Header-based options (CSS selector targeting, timeouts, token budgets, response formats) give it more depth than the one-liner suggests.
It's a good fit for developers who need good-enough extraction with zero setup. The self-hostable Apache-2.0 branch also provides a useful escape hatch for teams that can't let data leave their infrastructure. Heavy crawling workloads, JavaScript-dense targets, or sites behind aggressive bot protection are better served by Firecrawl or a headless-browser stack — Reader is a page reader, not a crawler.
Availability: the hosted API works keyless at 20 requests/min; a free API key adds 10M tokens and 500 RPM, with prepaid token packs (billed in USD via Stripe, priced in the dashboard) shared across Jina's Reader, Embeddings, and Reranker APIs. The OSS build ships as ghcr.io/jina-ai/reader:oss. Jina AI has been part of Elastic since October 2025; the service and repo remain active — the open-source branch was re-synced with production in April 2026 and was still receiving commits as of May 2026.
What people say
Developers reach for Jina Reader because it has almost no friction. Prepend https://r.jina.ai/ to any URL and clean markdown comes back: no SDK, no account, no configuration. Hacker News threads about HTML-to-markdown conversion reliably surface someone recommending it as the neat free API that handles the job, and that word-of-mouth is how most people find it. The companion s.jina.ai endpoint takes it a step further: it runs a web search and returns the top five results already converted, grounding an LLM with fresh data in a single API call.
The complaints are consistent too. Sites behind Cloudflare and similar bot protection frequently fail — community reports describe extraction as dependable on ordinary pages but weaker behind heavy JavaScript or aggressive bot protection. Head-to-head comparisons with Firecrawl generally find Firecrawl's markdown cleaner on JavaScript-heavy sites, and Reader is strictly a single-page tool: multi-page crawling means writing your own loop on top. Keyless access is capped at 20 requests a minute, search requests cost a fixed 10,000 tokens minimum, and pack prices are only visible inside the billing dashboard, which some users find opaque.
It's worth remembering the GitHub repo is the open-source branch of the codebase, not the full hosted service. It omits the MongoDB-backed SaaS layer and runs stateless or bucket-cached, but it ships a complete Docker image with headless Chrome, LibreOffice conversion, and CJK fonts, and it was re-synchronised with production code in April 2026 — this is a maintained OSS build, not an abandoned mirror.
Context for 2026: Elastic completed its acquisition of Jina AI in October 2025, with founder Han Xiao becoming Elastic's VP of AI. Reader, the token-metered API, and the open releases have continued since, though tighter Elasticsearch integration is clearly the direction of travel.
Summary of public user & expert reviews, compiled by RECATOOLS.
About this listing
This entry was compiled from publicly available data including Jina Reader's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Jina Reader unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to Jina Reader directly →
Spotted something out of date? Suggest an update →
More in Code & Dev Tools
