New supported embedding model

haileytap · haileytap · commit 055dc60de50a · 2025-07-16T18:01:44.000-04:00
diff --git a/articles/search/search-get-started-portal-image-search.md b/articles/search/search-get-started-portal-image-search.md
@@ -7,7 +7,7 @@ ms.author: haileytapia
 ms.service: azure-ai-search
 ms.update-cycle: 90-days
 ms.topic: quickstart
-ms.date: 06/11/2025
+ms.date: 07/16/2025
 ms.custom:
   - references_regions
 ---
@@ -52,7 +52,7 @@ For content embedding, you can choose either image verbalization (followed by te
 | Method | Description | Supported models |
 |--|--|--|
 | Image verbalization | Uses an LLM to generate natural-language descriptions of images, and then uses an embedding model to vectorize plain text and verbalized images.<br><br>Requires an [Azure OpenAI resource](/azure/ai-services/openai/how-to/create-resource) <sup>1, 2</sup> or [Azure AI Foundry project](/azure/ai-foundry/how-to/create-projects).<br><br>For text vectorization, you can also use an [Azure AI services multi-service resource](/azure/ai-services/multi-service-resource#azure-ai-multi-services-resource-for-azure-ai-search-skills) <sup>3</sup> in a [supported region](cognitive-search-skill-vision-vectorize.md). | LLMs:<br>GPT-4o<br>GPT-4o-mini<br>phi-4 <sup>4</sup><br><br>Embedding models:<br>text-embedding-ada-002<br>text-embedding-3-small<br>text-embedding-3-large |
-| Multimodal embeddings | Uses an embedding model to directly vectorize both text and images.<br><br>Requires an [Azure AI Foundry project](/azure/ai-foundry/how-to/create-projects) or [Azure AI services multi-service resource](/azure/ai-services/multi-service-resource#azure-ai-multi-services-resource-for-azure-ai-search-skills) <sup>3</sup> in a [supported region](cognitive-search-skill-vision-vectorize.md). | Cohere-embed-v3-english<br>Cohere-embed-v3-multilingual |
+| Multimodal embeddings | Uses an embedding model to directly vectorize both text and images.<br><br>Requires an [Azure AI Foundry project](/azure/ai-foundry/how-to/create-projects) or [Azure AI services multi-service resource](/azure/ai-services/multi-service-resource#azure-ai-multi-services-resource-for-azure-ai-search-skills) <sup>3</sup> in a [supported region](cognitive-search-skill-vision-vectorize.md). | Cohere-embed-v3-english<br>Cohere-embed-v3-multilingual<br>embed-v-4-0 <sup>5</sup> |
 
 <sup>1</sup> The endpoint of your Azure OpenAI resource must have a [custom subdomain](/azure/ai-services/cognitive-services-custom-subdomains), such as `https://my-unique-name.openai.azure.com`. If you created your resource in the [Azure portal](https://portal.azure.com/), this subdomain was automatically generated during resource setup.
 
@@ -62,6 +62,8 @@ For content embedding, you can choose either image verbalization (followed by te
 
 <sup>4</sup> `phi-4` is only available to Azure AI Foundry projects.
 
+<sup>5</sup> The Azure portal doesn't support `embed-v-4-0` for vectorization, so don't use it for this quickstart. Instead, use the [AML skill](cognitive-search-aml-skill.md) to programmatically specify this model. You can then use the portal to view and manage the skillset.
+
 ### Public endpoint requirements
 
 All of the preceding resources must have public access enabled so that the Azure portal nodes can access them. Otherwise, the wizard fails. After the wizard runs, you can enable firewalls and private endpoints on the integration components for security. For more information, see [Secure connections in the import wizards](search-import-data-portal.md#secure-connections).
diff --git a/articles/search/search-get-started-portal-import-vectors.md b/articles/search/search-get-started-portal-import-vectors.md
@@ -10,7 +10,7 @@ ms.custom:
   - build-2024
   - ignite-2024
 ms.topic: quickstart
-ms.date: 06/11/2025
+ms.date: 07/16/2025
 ---
 
 # Quickstart: Vectorize text in the Azure portal
@@ -49,7 +49,7 @@ For integrated vectorization, you must use one of the following embedding models
 |--|--|
 | [Azure OpenAI in Azure AI Foundry Models](/azure/ai-services/openai/how-to/create-resource) <sup>1, 2</sup> | text-embedding-ada-002<br>text-embedding-3-small<br>text-embedding-3-large |
 | [Azure AI services multi-service resource](/azure/ai-services/multi-service-resource#azure-ai-multi-services-resource-for-azure-ai-search-skills) <sup>3</sup> | For text and images: [Azure AI Vision multimodal](/azure/ai-services/computer-vision/how-to/image-retrieval) <sup>4</sup></li> |
-| [Azure AI Foundry model catalog](/azure/ai-foundry/what-is-azure-ai-foundry) | For text:<br>Cohere-embed-v3-english<br>Cohere-embed-v3-multilingual<br><br>For images:<br>Facebook-DinoV2-Image-Embeddings-ViT-Base<br>Facebook-DinoV2-Image-Embeddings-ViT-Giant |
+| [Azure AI Foundry model catalog](/azure/ai-foundry/what-is-azure-ai-foundry) | For text:<br>Cohere-embed-v3-english<br>Cohere-embed-v3-multilingual<br><br>For images:<br>Facebook-DinoV2-Image-Embeddings-ViT-Base<br>Facebook-DinoV2-Image-Embeddings-ViT-Giant<br><br>For text and images:<br>embed-v-4-0 <sup>5</sup> |
 
 <sup>1</sup> The endpoint of your Azure OpenAI resource must have a [custom subdomain](/azure/ai-services/cognitive-services-custom-subdomains), such as `https://my-unique-name.openai.azure.com`. If you created your resource in the [Azure portal](https://portal.azure.com/), this subdomain was automatically generated during resource setup.
 
@@ -59,6 +59,8 @@ For integrated vectorization, you must use one of the following embedding models
 
 <sup>4</sup> The Azure AI Vision multimodal embedding model is available in [select regions](/azure/ai-services/computer-vision/overview-image-analysis#region-availability).
 
+<sup>5</sup> The Azure portal doesn't support `embed-v-4-0` for vectorization, so don't use it for this quickstart. Instead, use the [AML skill](cognitive-search-aml-skill.md) to programmatically specify this model. You can then use the portal to view and manage the skillset.
+
 ### Public endpoint requirements
 
 For the purposes of this quickstart, all of the preceding resources must have public access enabled so that the Azure portal nodes can access them. Otherwise, the wizard fails. After the wizard runs, you can enable firewalls and private endpoints on the integration components for security. For more information, see [Secure connections in the import wizards](search-import-data-portal.md#secure-connections).
diff --git a/articles/search/search-how-to-integrated-vectorization.md b/articles/search/search-how-to-integrated-vectorization.md
@@ -7,7 +7,7 @@ author: haileytap
 ms.author: haileytapia
 ms.service: azure-ai-search
 ms.topic: how-to
-ms.date: 06/11/2025
+ms.date: 07/16/2025
 ---
 
 # Set up integrated vectorization in Azure AI Search using REST
@@ -48,7 +48,7 @@ For integrated vectorization, you must use one of the following embedding models
 |--|--|
 | [Azure OpenAI in Azure AI Foundry Models](/azure/ai-services/openai/how-to/create-resource) <sup>1, 2</sup> | text-embedding-ada-002<br>text-embedding-3-small<br>text-embedding-3-large |
 | [Azure AI services multi-service resource](/azure/ai-services/multi-service-resource#azure-ai-services-resource-for-azure-ai-search-skills) <sup>3</sup> | For text and images: [Azure AI Vision multimodal](/azure/ai-services/computer-vision/how-to/image-retrieval) <sup>4</sup></li> |
-<!--| [Azure AI Foundry model catalog](/azure/ai-foundry/what-is-azure-ai-foundry) | For text:<br>Cohere-embed-v3-english<br>Cohere-embed-v3-multilingual<br><br>For images:<br>Facebook-DinoV2-Image-Embeddings-ViT-Base<br>Facebook-DinoV2-Image-Embeddings-ViT-Giant |-->
+<!--| [Azure AI Foundry model catalog](/azure/ai-foundry/what-is-azure-ai-foundry) | For text:<br>Cohere-embed-v3-english<br>Cohere-embed-v3-multilingual<br><br>For images:<br>Facebook-DinoV2-Image-Embeddings-ViT-Base<br>Facebook-DinoV2-Image-Embeddings-ViT-Giant<br>For text and images:<br>embed-v-4-0|-->
 
 <sup>1</sup> The endpoint of your Azure OpenAI resource must have a [custom subdomain](/azure/ai-services/cognitive-services-custom-subdomains), such as `https://my-unique-name.openai.azure.com`. If you created your resource in the [Azure portal](https://portal.azure.com/), this subdomain was automatically generated during resource setup.
 
diff --git a/articles/search/tutorial-rag-build-solution-models.md b/articles/search/tutorial-rag-build-solution-models.md
@@ -9,7 +9,7 @@ ms.service: azure-ai-search
 ms.update-cycle: 180-days
 ms.topic: tutorial
 ms.custom: references_regions
-ms.date: 06/11/2025
+ms.date: 07/16/2025
 
 ---
 
@@ -54,11 +54,13 @@ Azure AI Search provides skill and vectorizer support for the following embeddin
 |--------|------------------|-------|------------|
 | Azure OpenAI | text-embedding-ada-002, <br>text-embedding-3-large, <br>text-embedding-3-small | [AzureOpenAIEmbedding](cognitive-search-skill-azure-openai-embedding.md) | [AzureOpenAIEmbedding](vector-search-vectorizer-azure-open-ai.md) |
 | Azure AI Vision | multimodal 4.0 <sup>1</sup> | [AzureAIVision](cognitive-search-skill-vision-vectorize.md) | [AzureAIVision](vector-search-vectorizer-ai-services-vision.md) |
-| Azure AI Foundry model catalog | Facebook-DinoV2-Image-Embeddings-ViT-Base, <br>Facebook-DinoV2-Image-Embeddings-ViT-Giant, <br>Cohere-embed-v3-english, <br>Cohere-embed-v3-multilingual | [AML](cognitive-search-aml-skill.md) <sup>2</sup>  | [Azure AI Foundry model catalog](vector-search-vectorizer-azure-machine-learning-ai-studio-catalog.md) |
+| Azure AI Foundry model catalog | Facebook-DinoV2-Image-Embeddings-ViT-Base, <br>Facebook-DinoV2-Image-Embeddings-ViT-Giant, <br>Cohere-embed-v3-english, <br>Cohere-embed-v3-multilingual, <br>embed-v-4-0 <sup>1, 2</sup> | [AML](cognitive-search-aml-skill.md) <sup>3</sup>  | [Azure AI Foundry model catalog](vector-search-vectorizer-azure-machine-learning-ai-studio-catalog.md) |
 
 <sup>1</sup> Supports image and text vectorization.
 
-<sup>2</sup> Deployed models in the model catalog are accessed over an AML endpoint. We use the existing AML skill for this connection.
+<sup>2</sup> At this time, you can only specify `embed-v-4-0` programmatically through the [AML skill](cognitive-search-aml-skill.md), not through the Azure portal. However, you can use the portal to view and manage the skillset afterward.
+
+<sup>3</sup> Deployed models in the model catalog are accessed over an AML endpoint. We use the existing AML skill for this connection.
 
 You can use other models besides the ones listed here. For more information, see [Use non-Azure models for embeddings](#use-non-azure-models-for-embeddings) in this article.
 
diff --git a/articles/search/vector-search-how-to-configure-vectorizer.md b/articles/search/vector-search-how-to-configure-vectorizer.md
@@ -10,7 +10,7 @@ ms.update-cycle: 180-days
 ms.custom:
   - build-2024
 ms.topic: how-to
-ms.date: 05/08/2025
+ms.date: 07/16/2025
 ---
 
 # Configure a vectorizer in a search index
@@ -44,10 +44,12 @@ The following table lists the embedding models that can be used with a vectorize
 | Vectorizer kind | Model names | Model provider | Associated skill |
 |-----------------|------------|----------------|------------------|
 | [`azureOpenAI`](vector-search-vectorizer-azure-open-ai.md) | text-embedding-ada-002, text-embedding-3 | Azure OpenAI | [AzureOpenAIEmbedding skill](cognitive-search-skill-azure-openai-embedding.md) |
-| [`aml`](vector-search-vectorizer-azure-machine-learning-ai-studio-catalog.md) | Facebook-DinoV2-Image-Embeddings, Cohere-embed-v3 | [Azure AI Foundry model catalog](vector-search-integrated-vectorization-ai-studio.md)  | [AML skill](cognitive-search-aml-skill.md) |
+| [`aml`](vector-search-vectorizer-azure-machine-learning-ai-studio-catalog.md) | Facebook-DinoV2-Image-Embeddings, Cohere-embed-v3, embed-v-4-0 <sup>1</sup> | [Azure AI Foundry model catalog](vector-search-integrated-vectorization-ai-studio.md)  | [AML skill](cognitive-search-aml-skill.md) |
 | [`aiServicesVision`](vector-search-vectorizer-ai-services-vision.md) | [Multimodal embeddings 4.0 API](/azure/ai-services/computer-vision/concept-image-retrieval) | Azure AI Vision (through an Azure AI services multi-service account) | [Azure AI Vision multimodal embeddings skill](cognitive-search-skill-vision-vectorize.md) |
 | [`customWebApi`](vector-search-vectorizer-custom-web-api.md) | Any embedding model | Hosted externally | [Custom Web API skill](cognitive-search-custom-skill-web-api.md) |
 
+<sup>1</sup> At this time, you can only specify `embed-v-4-0` programmatically through the [AML skill](cognitive-search-aml-skill.md), not through the Azure portal. However, you can use the portal to view and manage the skillset afterward.
+
 ## Try a vectorizer with sample data
 
 The [Import and vectorize data wizard](search-get-started-portal-import-vectors.md) reads files from Azure Blob storage, creates an index with chunked and vectorized fields, and adds a vectorizer. By design, the vectorizer that's created by the wizard is set to the same embedding model used to index the blob content.
diff --git a/articles/search/vector-search-integrated-vectorization-ai-studio.md b/articles/search/vector-search-integrated-vectorization-ai-studio.md
@@ -8,7 +8,7 @@ ms.service: azure-ai-search
 ms.custom:
   - build-2024
 ms.topic: how-to
-ms.date: 07/07/2025
+ms.date: 07/16/2025
 ---
 
 # Use embedding models from Azure AI Foundry model catalog for integrated vectorization
@@ -35,15 +35,13 @@ After the model is deployed, you can use it for [integrated vectorization](vecto
 
 Integrated vectorization and the [Import and vectorize data wizard](search-import-data-portal.md) support the following embedding models in the model catalog:
 
-For text embeddings:
+| Embedding type | Supported models |
+|--|--|
+| Text | Cohere-embed-v3-english, Cohere-embed-v3-multilingual|
+| Image | Facebook-DinoV2-Image-Embeddings-ViT-Base, Facebook-DinoV2-Image-Embeddings-ViT-Giant |
+| Multimodal (text and image) | embed-v-4-0 <sup>1</sup> |
 
-+ Cohere-embed-v3-english
-+ Cohere-embed-v3-multilingual
-
-For image embeddings:
-
-+ Facebook-DinoV2-Image-Embeddings-ViT-Base
-+ Facebook-DinoV2-Image-Embeddings-ViT-Giant
+<sup>1</sup> At this time, you can only specify `embed-v-4-0` programmatically through the [AML skill](cognitive-search-aml-skill.md), not through the Azure portal. However, you can use the portal to view and manage the skillset afterward.
 
 ## Deploy an embedding model from the Azure AI Foundry model catalog
 
@@ -178,15 +176,14 @@ This AML skill payload works with the following text embedding models from Azure
 
 + Cohere-embed-v3-english
 + Cohere-embed-v3-multilingual
++ embed-v-4-0
 
 It assumes that you're chunking your content using the Text Split skill and therefore your text to be vectorized is in the `/document/pages/*` path. If your text comes from a different path, update all references to the `/document/pages/*` path accordingly.
 
 You must add the `/v1/embed` path onto the end of the URL that you copied from your Azure AI Foundry deployment. You might also change the values for the `input_type`, `truncate` and `embedding_types` inputs to better fit your use case. For more information on the available options, review the [Cohere Embed API reference](/azure/ai-foundry/how-to/deploy-models-cohere-embed).
 
 The URI and key are generated when you deploy the model from the catalog. For more information about these values, see [How to deploy Cohere Embed models with Azure AI Foundry](/azure/ai-foundry/how-to/deploy-models-cohere-embed).
 
-Note that image URIs aren't supported by this integration at this time.
-
 ```json
 {
   "@odata.type": "#Microsoft.Skills.Custom.AmlSkill",
@@ -220,9 +217,9 @@ Note that image URIs aren't supported by this integration at this time.
 }
 ```
 
-In addition, the output of the Cohere model isn't the embeddings array directly, but rather a JSON object that contains it. You need to select it appropriately when mapping it to the index definition via `indexProjections` or `outputFieldMappings`. Here's a sample `indexProjections` payload that would allow you to do implement this mapping. 
+In addition, the output of the Cohere model isn't the embeddings array directly, but rather a JSON object that contains it. You need to select it appropriately when mapping it to the index definition via `indexProjections` or `outputFieldMappings`. Here's a sample `indexProjections` payload that would allow you to do implement this mapping.
 
-If you selected a different `embedding_types` in your skill definition that you have to change `float` in the `source` path to the appropriate type that you did select instead.
+If you selected a different `embedding_types` in your skill definition, change `float` in the `source` path to the type you selected.
 
 ```json
 "indexProjections": {
diff --git a/articles/search/vector-search-vectorizer-azure-machine-learning-ai-studio-catalog.md b/articles/search/vector-search-vectorizer-azure-machine-learning-ai-studio-catalog.md
@@ -8,13 +8,13 @@ ms.service: azure-ai-search
 ms.custom:
   - build-2024
 ms.topic: reference
-ms.date: 12/03/2024
+ms.date: 07/16/2024
 ---
 
 #	Azure AI Foundry model catalog vectorizer
 
 > [!IMPORTANT]
-> This vectorizer is in public preview under [Supplemental Terms of Use](https://azure.microsoft.com/support/legal/preview-supplemental-terms/). The [2024-05-01-Preview REST API](/rest/api/searchservice/indexes/create-or-update?view=rest-searchservice-2024-05-01-Preview&preserve-view=true) supports this feature.
+> This vectorizer is in public preview under [Supplemental Terms of Use](https://azure.microsoft.com/support/legal/preview-supplemental-terms/). To use this feature, we recommend the latest preview version of [Indexes - Create Or Update](/rest/api/searchservice/indexes/create-or-update) (REST API).
 
 The **Azure AI Foundry model catalog** vectorizer connects to an embedding model that was deployed via [the Azure AI Foundry model catalog](/azure/ai-foundry/how-to/model-catalog-overview) to an Azure Machine Learning endpoint. Your data is processed in the [Geo](https://azure.microsoft.com/explore/global-infrastructure/data-residency/) where your model is deployed. 
 
@@ -27,7 +27,7 @@ Parameters are case-sensitive. Which parameters you choose to use depends on wha
 | Parameter name | Description |
 |--------------------|-------------|
 | `uri` | (Required) The [URI of the AML online endpoint](../machine-learning/how-to-authenticate-online-endpoint.md) to which the _JSON_ payload is sent. Only the **https** URI scheme is allowed. |
-| `modelName` | (Required) The model ID from the Azure AI Foundry model catalog that is deployed at the provided endpoint. Supported models are: <ul><li>Facebook-DinoV2-Image-Embeddings-ViT-Base </li><li>Facebook-DinoV2-Image-Embeddings-ViT-Giant </li><li>Cohere-embed-v3-english </li><li>Cohere-embed-v3-multilingual</ul> |
+| `modelName` | (Required) The model ID from the Azure AI Foundry model catalog that is deployed at the provided endpoint. Supported models are:<p><ul><li>Facebook-DinoV2-Image-Embeddings-ViT-Base </li><li>Facebook-DinoV2-Image-Embeddings-ViT-Giant </li><li>Cohere-embed-v3-english </li><li>Cohere-embed-v3-multilingual</li><li>Cohere-embed-v4</li></ul> |
 | `key` | (Required for [key authentication](#WhatParametersToUse)) The [key for the AML online endpoint](../machine-learning/how-to-authenticate-online-endpoint.md). |
 | `resourceId` | (Required for [token authentication](#WhatParametersToUse)). The Azure Resource Manager resource ID of the AML online endpoint. It should be in the format subscriptions/{guid}/resourceGroups/{resource-group-name}/Microsoft.MachineLearningServices/workspaces/{workspace-name}/onlineendpoints/{endpoint_name}. |
 | `region` | (Optional for [token authentication](#WhatParametersToUse)). The [region](https://azure.microsoft.com/global-infrastructure/regions/) the AML online endpoint is deployed in. Needed if the region is different from the region of the search service. |
@@ -55,6 +55,7 @@ Which vector query types are supported by the Azure AI Foundry model catalog vec
 | Facebook-DinoV2-Image-Embeddings-ViT-Giant |  | X | X |
 | Cohere-embed-v3-english | X |  |  |
 | Cohere-embed-v3-multilingual | X |  |  |
+| Cohere-embed-v4 | X | X | X |
 
 ## Expected field dimensions
 
@@ -66,6 +67,7 @@ The expected field dimensions for a vector field configured with an Azure AI Fou
 | Facebook-DinoV2-Image-Embeddings-ViT-Giant | 1536 |
 | Cohere-embed-v3-english | 1024 |
 | Cohere-embed-v3-multilingual | 1024 |
+| Cohere-embed-v4 | 256–1536 |
 
 ## Sample definition
 
