Jina AI search models

Frontier-grade models for ingest and retrieval

Jina AI models preserve meaning from ingestion to retrieval. Get frontier accuracy at a fraction of the compute cost, deployable in your cloud, on-premises, or natively on Elasticsearch.

Try Jina on Elasticsearch
Learn more

AI models that preserve meaning from documents to answers

Jina AI develops frontier-grade small models for accurate, cost-effective document ingest, multilingual embedding across text and media, and reranking. It runs natively on Elasticsearch with zero config.

  • OCR/Reader

    Turn web pages, PDFs, scans, and handwriting into clean, structured input that search can retrieve, agents can act on, and retrieval augmented generation (RAG) pipelines ground their answers in.

    Get started

  • Embeddings

    Improve search and RAG systems with multimodal and multilingual embeddings for text, images, audio, video, and code.

    Get started

  • Reranker

    Maximize relevance with a world-class reranker that delivers precision for critical applications like RAG, AI assistant, and agents.

    Get started

Compact by design, precise by results

Turn documents into clean, structured data and high-precision search results, in one API.

  • Accurately ingest visual documents

    Turn documents and web content into clean, structured Markdown that keeps every number tied to what it means, so agents act on precise facts and RAG grounds answers in what content actually says.

  • Multimodal search, 100+ languages

    Jina's models work across text, images, audio, and video. With v5-omni, a single embedding model handles all four modalities in one shared space. Over 100 languages are supported natively, and cross-language search works out of the box.

  • Best results, not just nearest

    Jina's reranking models are proven leaders. Get extra precision with rerankers that rescore every candidate against the original query, using deep analysis to get the most relevant answers on top.

  • Smart training, smaller models

    Jina's models are trained on tasks that matter for retrieval: finding the right document and best answer from messy sources. That's why they match or outperform larger models at a fraction of the cost.

  • Map fields as semantic_text or semantic and Elasticsearch generates embeddings automatically. On EIS, Jina models default to deliver out-of-the-box multilingual and multimodal semantic search with zero config.

  • One API call, that's all

    Combine traditional keyword search with Jina's semantic matching in a single query. Use one API call with reciprocal rank fusion to merge the best of each approach.

Use Jina models wherever you build

From fully managed to self-hosted, Jina models meet you where your data lives. Pick the access path that fits.

Our research

Jina's models are built on research presented at top machine learning (ML) conferences, including CVPR, NeurIPS, and EMNLP. Explore how our frontier search models were trained from scratch in our latest publications.

  • Jina-embeddings-v5-text: Task-Targeted Embedding Distillation

    We introduce a novel training regimen that combines model distillation techniques with task-specific contrastive loss to produce compact, high-performance embedding models.

    Read more

  • jina-embeddings-v5-omni

    We extend jina-embeddings-v5-text to images, audio, and video by composing frozen pretrained encoders through lightweight trained adapters — without retraining the text model or reindexing existing data.

    Read more

  • jina-ranker-3.5

    We rework jina-reranker-v3's last-but-not-late listwise architecture with a hybrid sliding-window and global attention schedule, trained via multi-domain self-distillation.

    Read more

  • jina-ocr-v1

    We pair a compressed-vision encoder and MoE decoder with a FastMTP speculative decoding head sharing one draft block across three steps, trained via GRPO under verifiable rewards.

    Read more

Join our open source community

Jina's models are open-weight and freely available on Hugging Face, with millions of monthly downloads. The codebase is public on GitHub. The community has direct access to our developers.

Frequently asked questions

What are Jina search models?

Jina models are open source, frontier AI models for retrieval. They include embedding models for vectors, rerankers for precision, and OCR and readers for extracting and structuring content from URLs and docs.

Do I need AI or machine learning expertise to use Jina models?

ML expertise is not required. Jina models are available as preconfigured endpoints in Elasticsearch. Map a semantic or semantic_text field, and embedding happens automatically.

How do I get started?

Jina models are available on Elastic Inference Service on Elastic Cloud included in all trials. Start with defining a semantic or semantic_text field, or explore model cards on Hugging Face for code examples.

Which Jina models are available today?

Supported Jina models in Elastic Inference Service are detailed in our Jina models documentation.

How many languages are supported?

Jina-embeddings-v5-omni and Jina-ocr-v1 support 100+ languages. A query in one language finds relevant content written in another, with no translation pipelines required.

How does this relate to ELSER?

Jina embedding models are the default choice for new projects. ELSER will continue to be supported for customers that choose to use it. Both are available within Elasticsearch's hybrid search framework.

Is this a separate product?

Jina search models on the Elastic Inference Service are available to all Elastic Cloud users with consumption-based pricing. No separate license, subscription, or API key required. Jina models can be purchased separately as part of Jina On-Prem.

How does this relate to Elastic's vector database page?

The vector database page covers how vectors are stored and searched at scale. This page covers the AI models that generate and rerank them. Together: storage, compute, and application.

Try it now on Elastic Cloud

Spin up a fully loaded deployment on the cloud provider you choose. As the company behind Elasticsearch, we bring our features and support to your Elastic clusters in the cloud.

Start free trial