Merge pull request #7377 from MicrosoftDocs/main

learn-build-service-prod[bot] · web-flow · commit 9366ce79c9a2 · 2025-09-30T17:17:25.000Z
Auto Publish – main to live - 2025-09-30 17:13 UTC
diff --git a/articles/ai-foundry/agents/how-to/tools/model-context-protocol-samples.md b/articles/ai-foundry/agents/how-to/tools/model-context-protocol-samples.md
@@ -7,7 +7,7 @@ manager: nitinme
 ms.service: azure-ai-foundry
 ms.subservice: azure-ai-foundry-agent-service
 ms.topic: how-to
-ms.date: 09/04/2025
+ms.date: 09/30/2025
 author: aahill
 ms.author: aahi
 zone_pivot_groups: selection-mcp-code
@@ -16,9 +16,6 @@ ms.custom: azure-ai-agents-code
 
 # How to use the Model Context Protocol tool (preview)
 
-> [!NOTE]
-> Supported regions are `westus`, `westus2`, `uaenorth`, `southindia`, and `switzerlandnorth`.
-
 Use this article to find code samples for connecting Azure AI Foundry Agent Service with Model Context Protocol (MCP) servers.
 
 ## Prerequisites
diff --git a/articles/ai-foundry/agents/how-to/tools/model-context-protocol.md b/articles/ai-foundry/agents/how-to/tools/model-context-protocol.md
@@ -7,17 +7,13 @@ manager: nitinme
 ms.service: azure-ai-foundry
 ms.subservice: azure-ai-foundry-agent-service
 ms.topic: how-to
-ms.date: 09/04/2025
+ms.date: 09/30/2025
 author: aahill
 ms.author: aahi
-ms.custom: references_regions
 ---
 
 # Connect to Model Context Protocol servers (preview)
 
-> [!NOTE]
-> Supported regions are `westus`, `westus2`, `uaenorth`, `southindia`, and `switzerlandnorth`.
-
 > [!NOTE]
 > When using a [Network Secured Azure AI Foundry](../../how-to/virtual-networks.md), private MCP servers deployed in the same virtual network is not supported, only publicly accessible MCP servers are supported.
 
diff --git a/articles/ai-foundry/openai/concepts/prompt-engineering.md b/articles/ai-foundry/openai/concepts/prompt-engineering.md
@@ -4,7 +4,7 @@ titleSuffix: Azure OpenAI
 description: Learn how to use prompt engineering to optimize your work with Azure OpenAI.
 ms.service: azure-ai-openai
 ms.topic: conceptual
-ms.date: 09/23/2025
+ms.date: 09/30/2025
 ms.custom: references_regions, build-2023, build-2023-dataai
 manager: nitinme
 author: mrbullwinkle
@@ -102,27 +102,10 @@ Supporting content is information that the model can utilize to influence the ou
 
 ## Scenario-specific guidance
 
-While the principles of prompt engineering can be generalized across many different model types, certain models expect a specialized prompt structure. For Azure OpenAI GPT models, there are currently two distinct APIs where prompt engineering comes into play:
-
-- Chat Completion API.
-- Completion API.
-
-Each API requires input data to be formatted differently, which in turn impacts overall prompt design. The **Chat Completion API** supports the GPT-35-Turbo and GPT-4 models. These models are designed to take input formatted in a [specific chat-like transcript](../how-to/chatgpt.md) stored inside an array of dictionaries.
-
-The **Completion API** supports the older GPT-3 models and has much more flexible input requirements in that it takes a string of text with no specific format rules.
-
 The techniques in this section will teach you strategies for increasing the accuracy and grounding of responses you generate with a Large Language Model (LLM). It is, however, important to remember that even when using prompt engineering effectively you still need to validate the responses the models generate. Just because a carefully crafted prompt worked well for a particular scenario doesn't necessarily mean it will generalize more broadly to certain use cases. Understanding the [limitations of LLMs](/azure/ai-foundry/responsible-ai/openai/transparency-note#limitations), is just as important as understanding how to leverage their strengths.
 
-#### [Chat completion APIs](#tab/chat)
-
 [!INCLUDE [Prompt Chat Completion](../includes/prompt-chat-completion.md)]
 
-#### [Completion APIs](#tab/completion)
-
-[!INCLUDE [Prompt Completion](../includes/prompt-completion.md)]
-
----
-
 ## Best practices
 
 - **Be Specific**. Leave as little to interpretation as possible. Restrict the operational space.
@@ -133,7 +116,7 @@ The techniques in this section will teach you strategies for increasing the accu
 
 ## Space efficiency
 
-While the input size increases with each new generation of GPT models, there will continue to be scenarios that provide more data than the model can handle. GPT models break words into "tokens." While common multi-syllable words are often a single token, less common words are broken in syllables. Tokens can sometimes be counter-intuitive, as shown by the example below which demonstrates token boundaries for different date formats. In this case, spelling out the entire month is more space efficient than a fully numeric date. The current range of token support goes from 2,000 tokens with earlier GPT-3 models to up to 32,768 tokens with the 32k version of the latest GPT-4 model.
+While the input size increases with each new generation of GPT models, there will continue to be scenarios that provide more data than the model can handle. GPT models break words into "tokens." While common multi-syllable words are often a single token, less common words are broken in syllables. Tokens can sometimes be counter-intuitive, as shown by the example below which demonstrates token boundaries for different date formats. In this case, spelling out the entire month is more space efficient than a fully numeric date. 
 
 :::image type="content" source="../media/prompt-engineering/space-efficiency.png" alt-text="Screenshot of a string of text with highlighted colors delineating token boundaries." lightbox="../media/prompt-engineering/space-efficiency.png":::
 
diff --git a/articles/ai-foundry/openai/how-to/fine-tuning.md b/articles/ai-foundry/openai/how-to/fine-tuning.md
@@ -6,7 +6,7 @@ manager: nitinme
 ms.service: azure-ai-openai
 ms.custom: build-2023, build-2023-dataai, devx-track-python, references_regions
 ms.topic: how-to
-ms.date: 07/02/2025
+ms.date: 09/30/2025
 author: mrbullwinkle
 ms.author: mbullwin
 zone_pivot_groups: openai-fine-tuning
diff --git a/articles/ai-foundry/openai/how-to/function-calling.md b/articles/ai-foundry/openai/how-to/function-calling.md
@@ -7,7 +7,7 @@ ms.author: mbullwin #delegenz
 ms.service: azure-ai-openai
 ms.custom: devx-track-python
 ms.topic: how-to
-ms.date: 09/15/2025
+ms.date: 09/30/2025
 manager: nitinme
 ---
 
@@ -31,9 +31,6 @@ At a high level you can break down working with functions into three steps:
 
 * `gpt-35-turbo` (`1106`)
 * `gpt-35-turbo` (`0125`)
-* `gpt-4` (`1106-Preview`)
-* `gpt-4` (`0125-Preview`)
-* `gpt-4` (`vision-preview`)
 * `gpt-4` (`2024-04-09`)
 * `gpt-4o` (`2024-05-13`)
 * `gpt-4o` (`2024-08-06`)
@@ -44,6 +41,7 @@ At a high level you can break down working with functions into three steps:
 * `gpt-5` (`2025-08-07`)
 * `gpt-5-mini` (`2025-08-07`)
 * `gpt-5-nano` (`2025-08-07`)
+* `gpt-5-codex` (`2025-09-11`)
 
 Support for parallel function was first added in API version [`2023-12-01-preview`](https://github.com/Azure/azure-rest-api-specs/blob/main/specification/cognitiveservices/data-plane/AzureOpenAI/inference/preview/2023-12-01-preview/inference.json)
 
diff --git a/articles/ai-foundry/openai/how-to/json-mode.md b/articles/ai-foundry/openai/how-to/json-mode.md
@@ -6,7 +6,7 @@ services: cognitive-services
 manager: nitinme
 ms.service: azure-ai-openai
 ms.topic: how-to
-ms.date: 07/02/2025
+ms.date: 09/30/2025
 author: mrbullwinkle
 ms.author: mbullwin
 recommendations: false
diff --git a/articles/ai-foundry/openai/how-to/predicted-outputs.md b/articles/ai-foundry/openai/how-to/predicted-outputs.md
@@ -6,7 +6,7 @@ services: cognitive-services
 manager: nitinme
 ms.service: azure-ai-openai
 ms.topic: how-to
-ms.date: 06/17/2025
+ms.date: 09/30/2025
 author: mrbullwinkle
 ms.author: mbullwin
 recommendations: false
diff --git a/articles/ai-foundry/openai/how-to/switching-endpoints.yml b/articles/ai-foundry/openai/how-to/switching-endpoints.yml
@@ -7,7 +7,7 @@ metadata:
   author: mrbullwinkle
   ms.author: mbullwin
   manager: nitinme
-  ms.date: 09/15/2025
+  ms.date: 09/30/2025
   ms.service: azure-ai-openai
   ms.topic: how-to
   ms.custom:
diff --git a/articles/ai-foundry/openai/includes/embeddings-powershell.md b/articles/ai-foundry/openai/includes/embeddings-powershell.md
@@ -3,7 +3,7 @@ services: cognitive-services
 manager: nitinme
 ms.service: azure-ai-openai
 ms.topic: include
-ms.date: 12/05/2023
+ms.date: 09/30/2025
 author: mrbullwinkle #noabenefraim
 ms.author: mbullwin
 ---
diff --git a/articles/ai-foundry/openai/includes/embeddings-python.md b/articles/ai-foundry/openai/includes/embeddings-python.md
@@ -3,7 +3,7 @@ services: cognitive-services
 manager: nitinme
 ms.service: azure-ai-openai
 ms.topic: include
-ms.date: 09/01/2025
+ms.date: 09/30/2025
 author: mrbullwinkle #noabenefraim
 ms.author: mbullwin
 ---
diff --git a/articles/ai-foundry/openai/tutorials/embeddings.md b/articles/ai-foundry/openai/tutorials/embeddings.md
@@ -5,7 +5,7 @@ description: Learn how to use Azure OpenAI's embeddings API for document search
 manager: nitinme
 ms.service: azure-ai-openai
 ms.topic: tutorial
-ms.date: 07/02/2025
+ms.date: 09/30/2025
 author: mrbullwinkle #noabenefraim
 ms.author: mbullwin
 zone_pivot_groups: "openai-embeddings"
diff --git a/articles/ai-foundry/responsible-ai/speech-service/voice-live/data-privacy-security.md b/articles/ai-foundry/responsible-ai/speech-service/voice-live/data-privacy-security.md
@@ -0,0 +1,69 @@
+---
+title: Data, privacy, and security for Voice live
+titleSuffix: Azure AI services
+description: This document details issues for data, privacy, and security for Voice live.
+author: PatrickFarley
+ms.author: pafarley
+manager: nitinme
+ms.service: azure-ai-speech
+ms.topic: article
+ms.date: 09/29/2025
+---
+
+# Data, privacy, and security for Azure AI Voice Live API
+
+[!INCLUDE [non-english-translation](../../includes/non-english-translation.md)]
+
+> [!NOTE]
+> This article is provided for informational purposes only and not for the purpose of providing legal advice. We strongly recommend seeking specialist legal advice when implementing Speech Services.
+
+This article provides details regarding how data provided by you to the Azure AI Voice Live API ("Voice Live API") is processed, used, and stored.  
+
+Voice Live API is a fully managed service designed to empower developers to securely build, deploy, and scale high-quality, and extensible speech to speech experience for their voice agents. With Voice Live API, developers can choose from a list of different natively supported language models like GPT-Realtime, GPT-4.1, GPT-4o and GPT-4o-mini; incorporate an agent they have built using the Azure AI Foundry Agent Service to give the agent speech-in and speech-out capabilities; or bring their own model of choice deployed in Azure AI Foundry. 
+
+Voice Live API stores and processes data to provide the service and to monitor for violations of the applicable [Product Terms](https://www.microsoft.com/licensing/terms/). See also [the Microsoft Products and Services Data Protection Addendum](https://aka.ms/DPA), which governs data processing by the Azure AI services, including Voice Live API. Voice Live API is an Azure service;[ learn more about applicable Azure compliance offerings](/compliance/regulatory/offering-home). 
+
+> [!IMPORTANT]
+> Your prompts (inputs), completions (outputs), and your training data: 
+>
+> - are NOT available to other customers. 
+> - are NOT available to OpenAI or other model providers. 
+> - are NOT used to improve OpenAI models or other model providers’ models. 
+> - are NOT used to train, retrain, or improve Azure OpenAI Service or Azure AI Speech foundation models. 
+> - are NOT used to improve any Microsoft or third-party products or services without your permission or instruction. 
+>
+> With Voice Live API, your fine-tuned speech models are available exclusively for your use.
+
+The language models provided with Voice Live API are operated by Microsoft as an Azure service. If you choose to bring your own agent created with [Azure AI Foundry Agent Service](/azure/ai-foundry/agents/overview) or bring your deployed model in [Azure AI Foundry Models](/azure/ai-foundry/concepts/foundry-models-overview) to Voice Live API, additional information on data, privacy, and security is available at [Data, privacy, and security for Azure AI Foundry Agent Service](/azure/ai-foundry/responsible-ai/agents/data-privacy-security) and [Data, privacy, and security for use of models through the model catalog in Azure AI Foundry](/azure/ai-foundry/how-to/concept-data-privacy).
+
+## What data does Azure AI Voice live API process?
+
+Voice Live API processes the following types of data: 
+
+- **Prompts and output**. Prompts are submitted by the user, and content is generated by the GenAI model selected and converted to audio with or without avatar by Voice Live API. 
+- **Uploaded data**. You can provide your own data for use with Voice Live API using your own Azure Storage account or a configured data store, for example, your custom lexicon file to improve the pronunciation of the text to speech output, or your text and speech data to fine-tune the speech to text, text to speech and avatar model.  
+- **External data**. When you use the tools that support function calling, the service processes the outputs of those tools. 
+
+> [!IMPORTANT]
+> Custom neural voice ("custom voice") and custom avatar are available with [limited access](/azure/ai-foundry/responsible-ai/speech-service/text-to-speech/limited-access?tabs=cnv). Learn more about data processing, storage and retention for [custom text to speech (custom voice)](/azure/ai-foundry/responsible-ai/speech-service/text-to-speech/data-privacy-security?tabs=custom-neural-voice#recorded-acknowledgement-statement-verification) and [custom avatar](/azure/ai-foundry/responsible-ai/speech-service/text-to-speech/data-privacy-security?tabs=custom-avatar#video-acknowledgement-statement-verification).  
+
+## How does Azure AI Voice live API process data?
+
+The diagram below shows the data processing workflow for Voice Live API. It depicts how the API handles prompts (user audio input) through inferencing to produce content (agent audio output or video output with avatar), as well as how data from external tools is ingested into the service.
+
+:::image type="content" source="media/voice-live-diagram.png" alt-text="Diagram of the Voice live scenario.":::
+
+When these features are enabled by the user, Voice Live API processes audio input for noise suppression, echo cancellation, voice activity detection, and end of utterance detection , prior to sending the audio for speech recognition and language generation. For speech-to-speech models, audio output is generated directly from the language model. If a text-based language model is specified, Voice Live API converts the text response into audio. When an avatar is selected, the service streams the avatar and returns both the audio response and the avatar together.   
+
+When you bring your own model deployed in Azure AI Foundry or an agent built with Azure AI Foundry Agent Service to Voice Live API, the service interacts with the specified model endpoints to process your input prompts transcribed from audio and generate text output responses which may be further used or processed by Voice Live API for audio and avatar video generation. Data is processed for model inferencing in accordance with the terms that apply to the relevant model. Learn more at [Data, privacy, and security for Azure OpenAI Service](/azure/ai-foundry/responsible-ai/openai/data-privacy) and [Data, privacy, and security for use of models through the model catalog in AI Foundry portal](/azure/ai-studio/how-to/concept-data-privacy).
+
+To reduce the risk of harmful use of Voice Live API, the service includes [content filtering](/azure/ai-foundry/openai/concepts/content-filter) support. The outputs processed by the service will be filtered in accordance with any content filtering that has been applied to the natively supported models, or the model deployment used by your Foundry Agent. 
+
+
+## Data storage and retention  
+
+While Voice Live API itself does not store or retain customer data, the features (for example, custom voice, custom avatar, AI Foundry Agent) it interacts with may store customer data as the feature requires. Check data storage for [custom voice](/azure/ai-foundry/responsible-ai/speech-service/text-to-speech/data-privacy-security?tabs=custom-neural-voice#data-storage-and-retention), [custom avatar](/azure/ai-foundry/responsible-ai/speech-service/text-to-speech/data-privacy-security?tabs=custom-avatar#data-storage-and-retention), [AI Foundry Agents](/azure/ai-foundry/responsible-ai/agents/data-privacy-security#data-storage-for-azure-ai-agent-service-features), and [Azure OpenAI](/azure/ai-foundry/responsible-ai/openai/data-privacy?tabs=azure-portal#data-storage-for-azure-openai-service-features) if you are using these components. Learn more about [locations of processing for ‘global’ and ‘data zone’ deployments](/azure/ai-foundry/responsible-ai/openai/data-privacy?tabs=azure-portal#understanding-location-of-processing-for-global-and-data-zone-deployment-types).  
+
+Users can opt into a logging feature per debugging assistance from Microsoft engineers, when there is a [support ticket](/azure/ai-services/cognitive-services-support-options?context=%2Fazure%2Fai-services%2Fspeech-service%2Fcontext%2Fcontext#create-an-azure-support-request) filed. With this logging feature, users’ speech data is secured and stored in Azure storage managed by Microsoft within the same resource region. Microsoft’s debugging engineers are authorized Microsoft employees who access the data via point wise queries using request IDs, Secure Access Workstations (SAWs), and Just-In-Time (JIT) request approval granted by team managers. These logs are automatically removed in 30 days after generated.  
+
+To learn more about Microsoft's privacy and security commitments visit the [Microsoft Trust Center](https://www.microsoft.com/TrustCenter/CloudServices/Azure/default.aspx). 
diff --git a/articles/ai-foundry/responsible-ai/speech-service/voice-live/media/voice-live-diagram.png b/articles/ai-foundry/responsible-ai/speech-service/voice-live/media/voice-live-diagram.png
diff --git a/articles/ai-foundry/responsible-ai/speech-service/voice-live/transparency-note.md b/articles/ai-foundry/responsible-ai/speech-service/voice-live/transparency-note.md
diff --git a/articles/ai-services/speech-service/toc.yml b/articles/ai-services/speech-service/toc.yml
diff --git a/articles/ai-services/speech-service/voice-live-how-to-customize.md b/articles/ai-services/speech-service/voice-live-how-to-customize.md
diff --git a/articles/machine-learning/how-to-troubleshoot-environments.md b/articles/machine-learning/how-to-troubleshoot-environments.md
diff --git a/articles/machine-learning/reference-yaml-compute-instance.md b/articles/machine-learning/reference-yaml-compute-instance.md
