@@ -6,44 +6,47 @@ author: laujan
 manager: nitinme
 ms.service: azure-ai-language
 ms.topic: limits-and-quotas
-ms.date: 11/18/2025
+ms.date: 04/03/2026
 ms.author: lajanuar
 ---
+<!-- markdownlint-disable MD025 -->
 # Service limits for Azure Language in Foundry Tools
 
 > [!NOTE]
 > This article only describes the limits for preconfigured features in Azure Language in Foundry Tools:
-> To see the service limits for customizable features, see the following articles: 
+> To see the service limits for customizable features, see the following articles:
+>
 > * [Custom classification](../custom-text-classification/service-limits.md)
 > * [Custom NER](../custom-named-entity-recognition/service-limits.md)
 > * [Conversational language understanding](../conversational-language-understanding/service-limits.md)
 > * [Question answering](../question-answering/concepts/limits.md)
 
-Use this article to find the limits for the size, and rates that you can send data to the following features of Azure Language. 
-* [Named Entity Recognition (NER)](../named-entity-recognition/overview.md) 
+Use this article to find the limits for the size, and rates that you can send data to the following features of Azure Language.
+>
+* [Named Entity Recognition (NER)](../named-entity-recognition/overview.md)
 * [Personally Identifiable Information (PII) detection](../personally-identifiable-information/overview.md)
-* [Key phrase extraction](../key-phrase-extraction/overview.md) 
-* [Entity linking](../entity-linking/overview.md)  
+* [Key phrase extraction](../key-phrase-extraction/overview.md)
+* [Entity linking](../entity-linking/overview.md)
 * [Text Analytics for health](../text-analytics-for-health/overview.md)
 * [Sentiment analysis and opinion mining](../sentiment-opinion-mining/overview.md)
 * [Language detection](../language-detection/overview.md)
 
 When using features of Azure Language, keep the following information in mind:
 
-* Pricing is independent of data or rate limits. Pricing is based on the number of text records you send to the API, and is subject to your Language resource's [pricing details](https://aka.ms/unifiedLanguagePricing).
-    * A text record is measured as 1000 characters. 
-* Data and rate limits are based on the number of documents you send to the API. If you need to analyze larger documents than the limit allows, you can break the text into smaller chunks of text before sending them to the API. 
-* A document is a single string of text characters.  
+* Pricing is independent of data or rate limits. Pricing is based on the number of text records you send to the API, and is subject to your Language resource's [pricing details](https://aka.ms/unifiedLanguagePricing). **A text record is measured as 1000 characters**.
+* Data and rate limits are based on the number of documents you send to the API. If you need to analyze larger documents than the limit allows, you can break the text into smaller chunks of text before sending them to the API.
+* A document is a single string of text characters.
 
 ## Maximum characters per document
 
-The following limit specifies the maximum number of characters that can be in a single document.
+| The following limit specifies the maximum number of characters that can be in a single document. |
 
 | Feature | Value |
-|------------------------|---------------|
-| Text Analytics for health | 125,000 characters as measured by [StringInfo.LengthInTextElements](/dotnet/api/system.globalization.stringinfo.lengthintextelements).  | 
+| ------------------------ | --------------- |
+| Text Analytics for health | 125,000 characters as measured by [StringInfo.LengthInTextElements](/dotnet/api/system.globalization.stringinfo.lengthintextelements). |
+| Conversation PII | ≤ 1,000 characters as measured by the [**Analyze Conversations** API](/rest/api/language/analyze-conversations/analyze-conversations/analyze-conversations) default conversation item length. |
 | All other preconfigured features (synchronous) | 5,120 as measured by [StringInfo.LengthInTextElements](/dotnet/api/system.globalization.stringinfo.lengthintextelements). If you need to submit larger documents, consider using the feature asynchronously. |
-| All other preconfigured features ([asynchronous](use-asynchronously.md))  | 125,000 characters across all submitted documents, as measured by [StringInfo.LengthInTextElements](/dotnet/api/system.globalization.stringinfo.lengthintextelements) (maximum of 25 documents). |
+| All other preconfigured features ([asynchronous](use-asynchronously.md)) | 125,000 characters across all submitted documents, as measured by [StringInfo.LengthInTextElements](/dotnet/api/system.globalization.stringinfo.lengthintextelements) (maximum of 25 documents). |
 
 If a document exceeds the character limit, the API behaves differently depending on how you're sending requests.
 
@@ -65,10 +68,10 @@ The following limit specifies the maximum size of documents contained in the ent
 
 Exceeding the following document limits generates an HTTP 400 error code.
 
-> [!NOTE] 
+> [!NOTE]
 > When sending asynchronous API requests, you can send a maximum of 25 documents per request.
 
-| Feature | Max Documents Per Request | 
+| Feature | Max Documents Per Request |
 |----------|-----------|
 | Conversation summarization | 1 |
 | Language Detection | 1000 |

@@ -6,7 +6,7 @@ author: laujan
 manager: nitinme
 ms.service: azure-ai-language
 ms.topic: upgrade-and-migration-article
-ms.date: 02/05/2026
+ms.date: 04/03/2026
 ms.author: lajanuar
 ---
 <!-- markdownlint-disable MD025 -->
@@ -39,12 +39,8 @@ You can migrate to Microsoft Foundry using one of two approaches:
 
 * **[Option II: Migrate to a new Foundry resource](#option-ii-migrate-to-a-new-foundry-resource)**. Export your projects from Language Studio and import them into a new Foundry resource. This approach consolidates your AI capabilities into a single resource and provides access to both Foundry classic and the new Foundry experiences.
 
-
-
 ---
 
-
-
 ## Option I: Start using Foundry with an existing Language resource
 
 If you have an existing Azure Language resource with custom projects, you can continue using it within Microsoft Foundry by creating a Foundry hub and connecting your resource. This approach preserves your current resource configuration and allows you to access your projects in Foundry without exporting or reimporting data.
@@ -166,17 +162,22 @@ You can now train and deploy your `CNER` project using the **Getting started** w
 
 #### Import a Custom Question Answering (CQA) project
 
-1. In Foundry, navigate to your project.
-1. Select **Fine-tuning** from the left navigation pane.
-1. From the main window, select the **AI Service fine-tuning** tab, and then select **+ Fine-tune**.
-1. In the **Create CQA fine-tuning task** window, select your connected Azure AI Search resource.
-1. Enter a **Name** for your project and select the **Language**.
-1. Optionally, update the **Default answer when no answer is returned** field (default is "No answer found").
-1. Select **Create**.
-1. From the **Getting started** menu, select **Manage sources**.
-1. Select **+ Add source**, and then select **Add Files** to upload your exported question-answer pairs.
+1. In Foundry, navigate to your AI foundry project.
+1. When you export a `CQA` project from Language Studio, you receive a `.zip` file containing your project data. Extract the contents of the `.zip` file to access the underlying files.
+1. In Foundry, select **Fine-tuning** from the left navigation pane.
+1. From the main window, select the **AI Service fine-tuning** tab, and then select your existing `CQA` project or select **+ Fine-tune** to create a new fine-tuning task.
+1. On your `CQA` project Getting started page, select **Edit knowledge base**.
+1. On the **Edit knowledge base** page, select the **Question Answer Pairs** tab then select **Import question and answers**.
+1. Browse for your extracted files then drag and drop the `.tsv` file to import your question-answer pairs (If you had any question-answer pairs in the original project).
+
+    :::image type="content" source="media/question-answer-pairs-import.png" alt-text="Screenshot of the import question-answer pairs dialogue box window.":::
+
+1. Next, select the **Synonyms** tab on the page.
+1. Select **Import Synonyms** and choose the extracted `Synonyms.tsv` file (If you had any synonyms in the original project).
+
+    :::image type="content" source="media/synonyms-import.png" alt-text="Screenshot of the import synonyms dialogue box window.":::
 
-After adding your source files, you can train and deploy the `CQA` project using the **Getting started** workflow in Foundry.
+1. After adding your source files, you can train and deploy the `CQA` project using the **Getting started** workflow in Foundry.
 
 #### Import a Conversational Language Understanding (CLU) project
 

@@ -6,7 +6,7 @@ author: laujan
 manager: nitinme
 ms.service: azure-ai-language
 ms.topic: how-to
-ms.date: 11/18/2025
+ms.date: 04/03/2026
 ms.author: lajanuar
 ms.custom: language-service-pii
 ---
@@ -40,10 +40,6 @@ The API attempts to detect all the [defined entity categories](../concepts/conve
 
 For spoken transcripts, the entities detected are returned on the `redactionSource` parameter value provided. Currently, the supported values for `redactionSource` are `text`, `lexical`, `itn`, and `maskedItn` (which maps to Speech to text REST API's `display`\\`displayText`, `lexical`, `itn`, and `maskedItn` format respectively). Additionally, for the spoken transcript input, this API also provides audio timing information to empower audio redaction. For using the audioRedaction feature, use the optional `includeAudioRedaction` flag with `true` value. The audio redaction is performed based on the lexical input format.
 
-> [!NOTE]
-> Conversation PII now supports 40,000 characters as document size.
-
-
 ## Getting PII results
 
 When you get results from PII detection, you can stream the results to an application or save the output to a file on the local system. The API response includes [recognized entities](../concepts/conversations-entity-categories.md), including their categories and subcategories, and confidence scores. The text string with the PII entities redacted is also returned.