Jina AI

Search infrastructure: Reader (r.jina.ai turns any URL to markdown for grounding), plus multimodal/multilingual embeddings, reranker, DeepSearch, and Elasticsearch-native inference.

Jina AI offers (search/RAG infrastructure APIs (reader, web search, embeddings, reranker)) — ~4,000 max concurrent requests; avg latency 7.9s Reader / 2.5s Search.

grounding LLMs; RAG systems; search, RAG and agent systems; document analysis; improve factuality and reduce hallucinations

The data is delivered by api, dashboard. The supplier states the update frequency as "on-demand live fetch".

Coverage spans US; Application Software, Internet Services & Infrastructure.

Jina AI's site offers a live sample endpoint, and the supplier's own access route is an open api. Request a sample through Brickroad to start.

Jina AI publishes its data under CC BY-NC 4.0 (ReaderLM-v2 and jina-vlm models); Jina On-Prem commercial license; freemium — Reader keyless free at 20 RPM; every new API key = 10M free tokens at 500 RPM; premium key 5,000 RPM; token-metered billing (search costs fixed tokens from 10,000/request; ReaderLM-v2 costs 3x tokens); no dollar price points on fetched pages (/pricing 404). Terms and price bands are in the Buyer room.