Jina

Search + read-to-markdown T2 Search /jina

Jina specializes in turning the messy live web into clean, model-ready input. Its Reader takes any URL and returns pristine markdown with the boilerplate stripped away, and its search endpoint goes a step further — running a query and returning the full page content of the results in a single call. For RAG pipelines and agents that need trustworthy text rather than raw HTML, Jina is among the best tools for the job, and its embeddings and reranking models round out a complete retrieval stack.

Entry price
~token-based
MCP
official
Latency
low RPM

Choose if

You want clean LLM-ready markdown from a URL or a search-plus-read in a single call.

Avoid if

You need high search RPM or many results per query (≤5 default, low RPM).

Evals

How Jina performed on the curator-fleet web-access evals — the real measured numbers behind each composite, ranked best-first.

Parse · PDF/doccomposite 8.43/10 #2 of 3
0.78
Text accuracy
263 ms
Latency p50
$0 / useful
Cost
0%
Errors
Returns empty on scanned PDFs (no OCR).
Searchcomposite 7.69/10 Top pick#1 of 11
0.60
Recall@10
1.4 s
Latency p50
$0.00033 / useful result
Cost
0%
Errors
Wins all three query-classes.
Answercomposite 6.61/10 #7 of 7
1.00
Correctness
0.11
Citation faithfulness
2.6 s
Latency p50
$0.02 / call
Cost
0%
Errors
← All tools