Jina AI

Read access to the web and to Jina's own models is what an agent gets as tools: Reader turns a URL into markdown, s.jina.ai returns search results, and the embedding, reranking and classification APIs are reachable directly or through the MCP server at mcp.jina.ai. That is read only retrieval, with no actions in other systems and no per user credentials to scope.

Sourcejina.airead 2026-09-22

Each call to Jina's API stands alone: individual model calls (embeddings, rerank, classify, batch jobs and chat completions) and Reader requests run independently. No way to sequence, branch or route steps into a workflow is documented; pipelines that chain Reader, embeddings and reranking are built in the customer's own code or tools.

Sourceapi.jina.ai/openapi.jsonread 2026-09-22

A retrieval layer over the customer's own documents can be built from Jina's models: multimodal, multilingual embeddings, rerankers, a classifier and a segmenter for chunking, with batch embedding jobs and fine tuning of classifiers on the customer's data. Jina maintains no index or store of the customer's corpus; the vectors live in the customer's own database or Elasticsearch. The web search and Reader work over the web rather than the customer's corpus.

Sourcejina.ai/models/llms.txtread 2026-09-22

No approval step, review queue, escalation rule or policy constraint on what an agent does is documented. Jina's APIs read, embed and rank content and take no actions in other systems, and a per request token budget is a cost control.

Sourcejina.airead 2026-09-22

Jina's security statement describes customer isolation in dedicated trust zones, encryption at rest and in transit with per customer keys, internal and external penetration testing, secure development training and design reviews, and its terms offer a GDPR data processing agreement; a per request option keeps processing inside the EU. Access is by API key, and no SSO, roles or published attestation are documented.

Sourcejina.ai/legalread 2026-09-22

The API dashboard shows a key's available tokens and usage and publishes rate limits by tier; no per request log, trace or audit record of what an agent sent and received is documented. Token balances and rate limits are billing views, not a way to inspect prompts, tool calls and outputs step by step.

Sourcejina.ai/api-dashboard/pricingread 2026-09-22

No agent context carries between calls to Jina's APIs; stored classifiers and batch job outputs are model artifacts and results, not memory an agent draws on. No memory layer with a stated scope and lifetime is documented.

Sourceapi.jina.ai/openapi.jsonread 2026-09-22

Models run where the customer chooses. Jina's model guide says every model can be deployed fully air gapped in private infrastructure with no external network dependency, its pricing page offers deployment through the customer's own AWS or Azure account with billing through the cloud provider, and the hosted API has a per request option that keeps infrastructure and data processing inside the EU.

Sourcejina.ai/models/llms.txtread 2026-09-22

What Jina publishes is models (embeddings, rerankers, readers, OCR) and a guide to choosing among them: components a buyer builds with, not ready made workflows, templates or role specific agents.

Sourcejina.ai/models/llms.txtread 2026-09-22

Requests are answered as they arrive; no schedule, webhook, event subscription or channel that starts an agent is documented. A request and response API has nothing that wakes an agent without a person initiating the work.

Sourceapi.jina.ai/openapi.jsonread 2026-09-22

Whatever model the customer's agent runs can take retrieval from Jina, through its API and an MCP server any client can add. Within its own API the customer chooses among many embedding, reranking and reader models by setting the model parameter, including an OpenAI compatible chat completions endpoint. Model choice rests with the customer.

Sourcejina.ai/models/llms.txtread 2026-09-22

A documented API with a published OpenAPI specification (embeddings, rerank, classify, train, models, batch jobs and OpenAI compatible chat completions) makes Jina callable from outside, alongside the Reader and search endpoints at r.jina.ai and s.jina.ai, an MCP server at mcp.jina.ai, a CLI, and llms.txt guides for coding agents.

Sourceapi.jina.ai/openapi.jsonread 2026-09-22

Jina publishes research papers and benchmark results for its own models, and a speed test for Reader; none of these tests or scores the customer's agent. No testing, scoring or quality gate for the customer's agent workflows is documented.

Sourcejina.ai/models/llms.txtread 2026-09-22

The Reader loads a page in a browser engine Jina runs, with options for viewport size, cookies for authenticated pages, waiting for a selector, removing elements and running custom JavaScript on the page before it extracts the content as markdown. That is a headless browser fetch for extraction; an agent cannot use it to click through or operate an interface step by step.

Sourcejina.airead 2026-09-22