<?xml version="1.0" encoding="UTF-8"?>
<rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0">
  <channel>
    <title><![CDATA[Gregory Chase - Elasticsearch Labs]]></title>
    <description><![CDATA[Articles and tutorials from the Search team at Elastic]]></description>
    <copyright><![CDATA[© 2026. Elasticsearch B.V. All Rights Reserved]]></copyright>
    <image>
      <title><![CDATA[Gregory Chase - Elasticsearch Labs]]></title>
      <url>https://static-www.elastic.co/v3/assets/bltefdd0b53724fa2ce/blt1121c0bf0e8a6e65/6a88da6340a1841030ef456f/search-labs-thumbnail.png</url>
      <link>https://www.elastic.co/search-labs/author/gregory-chase</link>
    </image>
    <link>https://www.elastic.co/search-labs/author/gregory-chase</link>
    <atom:link href="https://www.elastic.co/search-labs/rss/author/gregory-chase.xml" rel="self" type="application/rss+xml"/>
    <language><![CDATA[en]]></language>
    <lastBuildDate>Fri, 20 Nov 2026 16:15:15 GMT</lastBuildDate>
  <item>
    <title><![CDATA[4 modalities, 100+ languages, 1 vector space: Jina Embeddings v5 in Microsoft Foundry]]></title>
    <description><![CDATA[Run the models on managed compute in your own Azure subscription, billed by compute hour, where nano indexes 3.4x faster than the more accurate small models.]]></description>
    <content:encoded><![CDATA[<p>One query can return a video clip and a paragraph in the same result set. Jina Embeddings v5 is now in the Microsoft Foundry model catalog, with four models between 239 million and 1.74 billion parameters. The <a href="https://jina.ai/news/jina-embeddings-v5-text-distilling-4b-quality-into-sub-1b-multilingual-embeddings/">text models</a> cover more than 100 languages, and the <a href="https://jina.ai/news/jina-embeddings-v5-omni-multimodal-embeddings-for-text-image-audio-and-video/">omni models</a> put text, images, PDFs, audio, and video into one vector space. </p><p></p><p>You deploy them as dedicated endpoints in your own Azure subscription and pay by compute hour; the model and your data stay in your tenant. The small models score 58.00 on the Massive Image Embedding Benchmark (MIEB) and 67.00 on the Massive Multilingual Text Embedding Benchmark (MMTEB); nano gives up about a point and a half on text and indexes 3.4x faster. Jina AI is part of Elastic, so the vectors go straight into Elasticsearch.</p><p></p><h2>What’s in the Jina Embeddings v5 model family</h2><p>Jina AI provides small, purpose-built models that preserve meaning across the entire search and retrieval process, delivering frontier-grade accuracy without the high costs. Jina models rank at or near the top of benchmarks, despite being a fraction of the size of their peers. In Foundry, Jina Embeddings v5 models bring highly relevant semantic search at lower compute costs. For full results, see the <a href="https://arxiv.org/abs/2602.15547">text</a> and<a href="https://arxiv.org/abs/2605.08384"> omni</a> technical reports.</p><p></p><p>The Jina Embeddings v5 family includes two types of models, each available in two parameter sizes:</p><p></p><ul><li><p><strong>Text models (</strong><strong><code>jina-embeddings-v5-text-small</code></strong><strong>, </strong><strong><code>jina-embeddings-v5-text-nano</code></strong><strong>).</strong> Multilingual text embeddings for retrieval, matching, clustering, and classification.</p></li></ul><p></p><ul><li><p><strong>Omni models </strong><strong>(</strong><strong><code>jina-embeddings-v5-omni-small</code></strong><strong>, </strong><strong><code>jina-embeddings-v5-omni-nano</code></strong><strong>).</strong> Add support for images, audio, and video, all in one vector space, enabling unified cross-modal search with a single query.</p></li></ul><p>Jina Embeddings v5 text models bring high accuracy in a small footprint across more than 100 languages. Omni models embed all modalities in the same vector space so a single query can return highly relevant results across all covered media and languages.</p><h3>How the small and nano models compare on accuracy and speed</h3><p>Choose the small models when accuracy is your most important requirement. They have larger context windows at 32,000 tokens and more parameters to maximize the relevance of the content that they embed and retrieve. </p><p></p><p>When speed and throughput are more important, choose the nano models. The following benchmarks illustrate this tradeoff. </p><p></p><p>For multimodal applications, v5-omni-small scores 58.00 on MIEB (image) and 41.20 on the Massive Multimodal Embedding Benchmark–Video (MMEB-Video) subset; whereas v5-omni-nano scores 47.87 and 26.87, respectively.  </p><p></p><p>Both models have similar performance for text (small scores 67.00 on MMTEB; and nano scores 65.5 on MMTEB) and audio (both score 49.69 on the Massive Audio Embedding Benchmark [MAEB]). For throughput benchmarks, we see nano models processing documents 3.4x faster for text, 2.2x faster for images and video, and 1.5x faster for audio than small models.</p><p></p><p><strong>Model</strong></p><p><strong>Input</strong></p><p><strong>Parameters</strong></p><p><strong>Input token length</strong></p><p><strong>Matryoshka dimensions</strong></p><p><strong>Best for</strong></p><p><code>jina-embeddings-v5-text-small</code></p><p>Text</p><p>677 million</p><p>32,000</p><p>1024–32</p><p>Highest semantic search accuracy in a small footprint</p><p><code>jina-embeddings-v5-text-nano</code></p><p>Text</p><p>239 million</p><p>8,000</p><p>768–32</p><p>Up to 3.4x faster document indexing and retrieval in a tiny footprint</p><p><code>jina-embeddings-v5-omni-small</code></p><p>Text, image, PDF, audio, video</p><p>~1.74 billion </p><p>32,000</p><p>1024–32</p><p>Highest semantic search accuracy, incorporating text, images, video, and audio</p><p><code>jina-embeddings-v5-omni-nano</code></p><p>Text, image, PDF, audio, video</p><p>~1.04 billion </p><p>8,000</p><p>768–32</p><p>1.5–3.4x faster indexing and retrieval of multimodal content</p><p><em>Table 1. Differences between the variants of the </em><em><code>jina-embeddings-v5</code></em><em> model family.</em></p><h2>How to deploy Jina Embeddings v5 in Microsoft Foundry</h2><h3>Provision a Jina Embeddings v5 endpoint on managed compute</h3><ol><li><p>Log in to your Azure account. </p></li><li><p>Open the <strong>Microsoft Foundry</strong> model catalog, and locate <strong>Jina Embeddings v5</strong>.</p></li><li><p>Select your model variant, and deploy it to managed compute. Choose a supported GPU SKU (the model cards list A10), and make sure your Azure subscription has the required quota for that SKU. Set the number of instances. (Each one runs its own copy of the model, so add instances for more traffic.) </p></li><li><p>Select <strong>Deploy</strong>.</p></li><li><p>Your endpoint provisions in 15 to 20 minutes.</p></li></ol><p></p><h3>Call your Jina Embeddings v5 endpoint</h3><p>In <strong>Models + endpoints</strong>, open your deployment and copy the <strong>Target URI</strong> and <strong>key</strong>. Use the URI as shown; it already ends in <code>/v1/embeddings</code>. The request body is the one that the Jina Embeddings API uses:</p>curl "$TARGET_URI" \
  -H "Authorization: Bearer $ENDPOINT_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "jina-embeddings-v5-text-small",
    "task": "retrieval.passage",
    "input": ["Invoices are due 30 days after delivery."]
  }'<p>When you call it:</p><p></p><ul><li><p>Set <code>task</code> to <code>retrieval.passage</code> when you index and to <code>retrieval.query</code> when you search.</p></li><li><p>With an omni model, pass media in the same list as text, as Base64; for example, <code>{"image": "&lt;base64-encoded image&gt;"}</code>.</p></li><li><p>Catalog deployments block outbound access, so media URLs aren’t supported. </p></li></ul><p></p><p>Then store the vectors in Elasticsearch or another vector store.</p><p></p><h2>Jina Embeddings v5 in the Microsoft Foundry model catalog</h2><p>For more information about the Jina models in the Microsoft Foundry model catalog, click the following links. Each model has its own catalog listing, with the model card, supported GPU SKUs, and pricing:</p><p></p><p></p><p></p><ul><li><p><a href="https://ai.azure.com/catalog/models/jina-embeddings-v5-text-small?publisher=elastic">jina-embeddings-v5-text-small</a></p></li><li><p><a href="https://ai.azure.com/catalog/models/jina-embeddings-v5-text-nano?publisher=elastic">Jina-embeddings-v5-text-nano</a></p></li><li><p><a href="https://ai.azure.com/catalog/models/jina-embeddings-v5-omni-small?publisher=elastic">jina-embeddings-v5-omni-small</a></p></li><li><p><a href="https://ai.azure.com/catalog/models/jina-embeddings-v5-omni-nano?publisher=elastic">jina-embeddings-v5-omni-nano</a></p></li></ul><p></p><h2>Jina Embeddings v5 model cards and technical reports</h2><ul><li><p>Model cards on Hugging Face: <a href="https://huggingface.co/jinaai/jina-embeddings-v5-text-small">text-small</a>, <a href="https://huggingface.co/jinaai/jina-embeddings-v5-text-nano">text-nano</a>, <a href="https://huggingface.co/jinaai/jina-embeddings-v5-omni-small">omni-small</a>, <a href="https://huggingface.co/jinaai/jina-embeddings-v5-omni-nano">omni-nano</a></p></li><li><p><a href="https://arxiv.org/abs/2605.08384">Jina omni technical report</a></p></li><li><p><a href="https://arxiv.org/abs/2602.15547">Jina text technical report</a></p></li></ul>]]></content:encoded>
    <link>https://www.elastic.co/search-labs/blog/jina-embeddings-v5-microsoft-foundry</link>
    <guid isPermaLink="true">https://www.elastic.co/search-labs/blog/jina-embeddings-v5-microsoft-foundry</guid>
    <category><![CDATA[Jina AI]]></category>
    <category><![CDATA[Vector Database]]></category>
    <category><![CDATA[Integrations]]></category>
    <dc:creator><![CDATA[Sa Zhang,Zac Li,Gregory Chase]]></dc:creator>
    <enclosure url="https://static-www.elastic.co/v3/assets/bltefdd0b53724fa2ce/blt1709fc0b3e294667/6b00503fdae41d00f9c75617/jina-embeddings-v5-microsoft-foundry.jpg" length="0" type="image/jpeg"/>
    <pubDate>Fri, 20 Nov 2026 16:00:00 GMT</pubDate>
  </item>
  </channel>
</rss>