DeepSeek-R1-0528 is now available on Azure AI Foundry
We’re excited to announce that DeepSeek-R1-0528, the latest evolution in the DeepSeek R1 open-source series of reasoning-optimized models, is now available on the Azure AI Foundry. According to DeepSeek, the R1-0528 model brings improved depth of reasoning and inferencing capabilities, and has demonstrated outstanding performance across various benchmark evaluations, approaching leading models such as OpenAI o3 and Gemini 2.5 Pro. In less than 36 hours, we’ve seen 4x growth in deployments of DeepSeek-R1-0528 compared to DeepSeek R1. Building on the foundation of DeepSeek-R1, this new release continues to push the boundaries of advanced reasoning and task decomposition. DeepSeek-R1-0528 integrates enhancements in chain-of-thought prompting, reinforcement learning fine-tuning, and broader multilingual understanding, making it a powerful tool for developers building intelligent agents, copilots, and research applications. Available within Azure AI Foundry, DeepSeek-R1-0528 is accessible on a trusted, scalable, and enterprise-ready platform, enabling businesses to seamlessly integrate advanced AI while meeting SLAs, security, and responsible AI commitments -all backed by Microsoft’s reliability and innovation. What’s new in DeepSeek-R1-0528? While maintaining the core strengths of its predecessor, DeepSeek-R1-0528 introduces: Improved reasoning depth through refined CoT (Chain-of-Thought) strategies. Expanded dataset coverage for better generalization across domains. Optimized inference performance for faster response times in production environments. New algorithmic optimization mechanisms during post-training. DeepSeek-R1-0528 is joining other direct from Azure models and it will be hosted and sold by Azure. Build Trustworthy AI Solutions with Azure AI Foundry As part of our ongoing commitment to help customers use and build AI that is trustworthy, meaning AI that is secure, safe and private, DeepSeek-R1-0528 has undergone Azure’s safety evaluations, including assessments of model behavior and automated security reviews to mitigate potential risks. With Azure AI Content Safety, built-in content filtering is available by default, with opt-out options for flexibility. We suggest using Azure AI Content Safety and conducting independent evaluations in production, as researchers have found DeepSeek-R1-0528 scoring lower than other models—though in line with DeepSeek-R1—on safety and jailbreak benchmarks. Get started today You can explore and deploy DeepSeek-R1-0528 directly from the Azure AI Foundry model catalog or integrate it into your workflows using the Azure AI SDK. The model is also available for experimentation via GitHub. Whether you're building a domain-specific assistant, a research prototype, or a production-grade AI system, DeepSeek-R1-0528 offers a robust foundation for your next breakthrough.Unlocking Document Intelligence: Mistral OCR Now Available in Azure AI Foundry
Every organization has a treasure trove of information—buried not in databases, but in documents. From scanned contracts and handwritten forms to research papers and regulatory filings, this knowledge often sits locked in static formats, invisible to modern AI systems. Imagine if we could teach machines not just to read, but to truly understand the structure and nuance of these documents. What if equations, images, tables, and multilingual text could be seamlessly extracted, indexed, and acted upon—at scale? That future is here. Today we are announcing the launch of Mistral OCR in the Azure AI Foundry model catalog—a state-of-the-art Optical Character Recognition (OCR) model that brings intelligent document understanding to a whole new level. Designed for speed, precision, and multilingual versatility, Mistral OCR unlocks the potential of unstructured content with unmatched performance. From Patient Charts to Investment Reports—Built for Every Industry Mistral OCR’s ability to extract structure from complex documents makes it transformative across a range of verticals: Healthcare Hospitals and health systems can digitize clinical notes, lab results, and patient intake forms, transforming scanned content into structured data for downstream AI applications—improving care coordination, automation, and insights. Finance & Insurance From loan applications and KYC documents to claims forms and regulatory disclosures, Mistral OCR helps financial institutions process sensitive documents faster, more accurately, and with multilingual support—ensuring compliance and improving operational efficiency. Education & Research Academic institutions and research teams can turn PDFs of scientific papers, course materials, and diagrams into AI-readable formats. Mistral OCR’s support for equations, charts, and LaTeX-style formatting makes it ideal for scientific knowledge extraction. Legal & Government With its multilingual and high-fidelity OCR capabilities, legal teams and public agencies can digitize contracts, historical records, and filings—accelerating review workflows, preserving archival materials, and enabling transparent governance. Key Highlights of Mistral OCR According to Mistral their OCR model stands apart due to the following: State-of-the-Art Document Understanding Mistral OCR excels in parsing complex, multimodal documents—extracting tables, math, and figures with markdown-style clarity. It goes beyond recognition to deliver understanding. benchmark testing. Whether you’re working in Hindi, Arabic, French, or Chinese—this model adapts seamlessly. State-of-the-Art Document Understanding Mistral OCR excels in parsing complex, multimodal documents—extracting tables, math, and figures with markdown-style clarity. It goes beyond recognition to deliver understanding. Multilingual by Design With support for dozens of languages and scripts, Mistral OCR achieves 99%+ fuzzy match scores in benchmark testing. Whether you’re working in Hindi, Arabic, French, or Chinese—this model adapts seamlessly. Fastest in Its Class Process up to 2,000 pages per minute on a single node. This speed makes it ideal for enterprise document pipelines and real-time applications. Doc-as-Prompt + Structured Output Turn documents into intelligent prompts—then extract structured, JSON-formatted output for downstream use in agents, workflows, or analytics engines. Why use Mistral OCR on Azure AI Foundry? Mistral OCR is now available as serverless APIs through Models as a Service (MaaS) in Azure AI Foundry. This enables enterprise-scale workloads with ease. Network Isolation for Inferencing: Protect your data from public network access. Expanded Regional Availability: Access from multiple regions. Data Privacy and Security: Robust measures to ensure data protection. Quick Endpoint Provisioning: Set up an OCR endpoint in Azure AI Foundry in seconds. Azure AI ensures seamless integration, enhanced security, and rapid deployment for your AI needs. How to deploy Mistral OCR model in Azure AI Foundry? Prerequisites: If you don’t have an Azure subscription, get one here: https://azure.microsoft.com/en-us/pricing/purchase-options/pay-as-you-go Familiarize yourself with Azure AI Model Catalog Create an Azure AI Foundry hub and project. Make sure you pick East US, West US3, South Central US, West US, North Central US, East US 2 or Sweden Central as the Azure region for the hub. Create a deployment to obtain the inference API and key: Open the model card in the model catalog on Azure AI Foundry. Click on Deploy and select the Pay-as-you-go option. Subscribe to the Marketplace offer and deploy. You can also review the API pricing at this step. You should land on the deployment page that shows you the API and key in less than a minute. These steps are outlined in detail in the product documentation. From Documents to Decisions The ability to extract meaning from documents—accurately, at scale, and across languages—is no longer a bottleneck. With Mistral OCR now available in Azure AI Foundry, organizations can move beyond basic text extraction to unlock true document intelligence. This isn’t just about reading documents. It’s about transforming how we interact with the knowledge they contain. Try it. Build with it. And see what becomes possible when documents speak your language.Integrate Custom Azure AI Agents with CoPilot Studio and M365 CoPilot
Integrating Custom Agents with Copilot Studio and M365 Copilot In today's fast-paced digital world, integrating custom agents with Copilot Studio and M365 Copilot can significantly enhance your company's digital presence and extend your CoPilot platform to your enterprise applications and data. This blog will guide you through the integration steps of bringing your custom Azure AI Agent Service within an Azure Function App, into a Copilot Studio solution and publishing it to M365 and Teams Applications. When Might This Be Necessary: Integrating custom agents with Copilot Studio and M365 Copilot is necessary when you want to extend customization to automate tasks, streamline processes, and provide better user experience for your end-users. This integration is particularly useful for organizations looking to streamline their AI Platform, extend out-of-the-box functionality, and leverage existing enterprise data and applications to optimize their operations. Custom agents built on Azure allow you to achieve greater customization and flexibility than using Copilot Studio agents alone. What You Will Need: To get started, you will need the following: Azure AI Foundry Azure OpenAI Service Copilot Studio Developer License Microsoft Teams Enterprise License M365 Copilot License Steps to Integrate Custom Agents: Create a Project in Azure AI Foundry: Navigate to Azure AI Foundry and create a project. Select 'Agents' from the 'Build and Customize' menu pane on the left side of the screen and click the blue button to create a new agent. Customize Your Agent: Your agent will automatically be assigned an Agent ID. Give your agent a name and assign the model your agent will use. Customize your agent with instructions: Add your knowledge source: You can connect to Azure AI Search, load files directly to your agent, link to Microsoft Fabric, or connect to third-party sources like Tripadvisor. In our example, we are only testing the CoPilot integration steps of the AI Agent, so we did not build out additional options of providing grounding knowledge or function calling here. Test Your Agent: Once you have created your agent, test it in the playground. If you are happy with it, you are ready to call the agent in an Azure Function. Create and Publish an Azure Function: Use the sample function code from the GitHub repository to call the Azure AI Project and Agent. Publish your Azure Function to make it available for integration. azure-ai-foundry-agent/function_app.py at main · azure-data-ai-hub/azure-ai-foundry-agent Connect your AI Agent to your Function: update the "AIProjectConnString" value to include your Project connection string from the project overview page of in the AI Foundry. Role Based Access Controls: We have to add a role for the function app on OpenAI service. Role-based access control for Azure OpenAI - Azure AI services | Microsoft Learn Enable Managed Identity on the Function App Grant "Cognitive Services OpenAI Contributor" role to the System-assigned managed identity to the Function App in the Azure OpenAI resource Grant "Azure AI Developer" role to the System-assigned managed identity for your Function App in the Azure AI Project resource from the AI Foundry Build a Flow in Power Platform: Before you begin, make sure you are working in the same environment you will use to create your CoPilot Studio agent. To get started, navigate to the Power Platform (https://make.powerapps.com) to build out a flow that connects your Copilot Studio solution to your Azure Function App. When creating a new flow, select 'Build an instant cloud flow' and trigger the flow using 'Run a flow from Copilot'. Add an HTTP action to call the Function using the URL and pass the message prompt from the end user with your URL. The output of your function is plain text, so you can pass the response from your Azure AI Agent directly to your Copilot Studio solution. Create Your Copilot Studio Agent: Navigate to Microsoft Copilot Studio and select 'Agents', then 'New Agent'. Make sure you are in the same environment you used to create your cloud flow. Now select ‘Create’ button at the top of the screen From the top menu, navigate to ‘Topics’ and ‘System’. We will open up the ‘Conversation boosting’ topic. When you first open the Conversation boosting topic, you will see a template of connected nodes. Delete all but the initial ‘Trigger’ node. Now we will rebuild the conversation boosting agent to call the Flow you built in the previous step. Select 'Add an Action' and then select the option for existing Power Automate flow. Pass the response from your Custom Agent to the end user and end the current topic. My existing Cloud Flow: Add action to connect to existing Cloud Flow: When this menu pops up, you should see the option to Run the flow you created. Here, mine does not have a very unique name, but you see my flow 'Run a flow from Copilot' as a Basic action menu item. If you do not see your cloud flow here add the flow to the default solution in the environment. Go to Solutions > select the All pill > Default Solution > then add the Cloud Flow you created to the solution. Then go back to Copilot Studio, refresh and the flow will be listed there. Now complete building out the conversation boosting topic: Make Agent Available in M365 Copilot: Navigate to the 'Channels' menu and select 'Teams + Microsoft 365'. Be sure to select the box to 'Make agent available in M365 Copilot'. Save and re-publish your Copilot Agent. It may take up to 24 hours for the Copilot Agent to appear in M365 Teams agents list. Once it has loaded, select the 'Get Agents' option from the side menu of Copilot and pin your Copilot Studio Agent to your featured agent list Now, you can chat with your custom Azure AI Agent, directly from M365 Copilot! Conclusion: By following these steps, you can successfully integrate custom Azure AI Agents with Copilot Studio and M365 Copilot, enhancing you’re the utility of your existing platform and improving operational efficiency. This integration allows you to automate tasks, streamline processes, and provide better user experience for your end-users. Give it a try! Curious of how to bring custom models from your AI Foundry to your CoPilot Studio solutions? Check out this blogAzure AI Foundry Models: Futureproof Your GenAI Applications
Years of Rapid Growth and Innovation The Azure AI Foundry Models journey started with the launch of Models as a Service (MaaS) in partnership with Meta Llama at Ignite 2023. Since then, we’ve rapidly expanded our catalog and capabilities: 2023: General Availability of the model catalog and launch of MaaS 2024: 1800+ models available including Cohere, Mistral, Meta, G42, AI21, Nixtla and more, with 250+ OSS models deployed on managed compute 2025 (Build): 10000+ models, new models sold directly by Microsoft, more managed compute models and expanded partnerships, introduction of advanced tooling like Model Leaderboard, Model Router, MCP Server, and Image Playground GenAI Trends Reshaping the Model Landscape To stay ahead of the curve, Azure AI Foundry Models is designed to support the most important trends in GenAI: Emergence of Reasoning-Centric Models Proliferation of Agentic AI and Multi-agent systems Expansion of Open-Source Ecosystems Multimodal Intelligence Becoming Mainstream Rise of Small, Efficient Models (SLMs) These trends are shaping a future where enterprises need not just access to models—but smart tools to pick, combine, and deploy the best ones for each task. A Platform Built for Flexibility and Scale Azure AI Foundry is more than a catalog—it’s your end-to-end platform for building with AI. You can: Explore over 10000+ models, including foundation, industry, multimodal, and reasoning models along with agents. Deploy using flexible options like PayGo, Managed Compute, or Provisioned Throughput (PTU) Monitor and optimize performance with integrated observability and compliance tooling Whether you're prototyping or scaling globally, Foundry gives you the flexibility you need. Two Core Model Categories 1. Models Sold Directly by Microsoft These models are hosted and billed directly by Microsoft under Microsoft Product Terms. They offer: Enterprise-grade SLAs and reliability Deep Azure service integration Responsible AI standards Flexible usage of reserved quota by using Azure AI Foundry Provisioned Throughput (PTU) across direct models including OpenAI, Meta, Mistral, Grok, DeepSeek and Black Forest Labs. Reduce AI workload costs on predictable consumption patterns with Azure AI Foundry Provisioned Throughput reservations. Learn more here Coming to the family of direct models from Azure: Grok 3 / Grok 3 Mini (from xAI) Flux Pro 1.1 Ultra (from Black Forest Labs) Llama 4 Scout & Maverick (from Meta) Codestral 2501, OCR (from Mistral) 2. Models from Partners & Community These models come from the broader ecosystem, including open-source and monetized partners. They are deployed as Managed Compute or Standard PayGo, and include models from Cohere, Paige and Saifr. We also have new industry models joining this ecosystem of partner and community models NVIDIA NIMs: ProteinMPNN, RFDiffusion, OpenFold2, MSA Paige AI: Virchow 2G, Virchow 2G-mini Microsoft Research: EvoDiff, BioEmu-1 Expanded capabilities that make model choice simpler and faster Azure AI Foundry Models isn’t just about more models. We’re introducing tools to help developers intelligently navigate model complexity: 1. Model Leaderboard Easily compare model performance across real-world tasks with: Transparent benchmark scores Task-specific rankings (summarization, RAG, classification, etc.) Live updates as new models are evaluated Whether you want the highest accuracy, fastest throughput, or best price-performance ratio—the leaderboard guides your selection. 2. Model Router Don’t pick just one—let Azure do the heavy lifting. Automatically route queries to the best available model Optimize based on speed, cost, or quality Supports dynamic fallback and load balancing This capability is a game-changer for agents, copilots, and apps that need adaptive intelligence. 3. Image/Video Playground A new visual interface for: Testing image generation models side-by-side Tuning prompts and decoding settings Evaluating output quality interactively This is particularly useful for multimodal experimentation across marketing, design, and research use cases. 4. MCP Server Enables model-aware orchestration, especially for agentic workloads: Tool use integration Multi-model planning and reasoning Unified coordination across model APIs A Futureproof Foundation With Azure AI Foundry Models, you're not just selecting from a list of models—you’re stepping into a full-stack, flexible, and future-ready AI environment: Choose the best model for your needs Deploy on your terms—serverless, managed, or reserved Rely on enterprise-grade performance, security, and governance Stay ahead with integrated innovation from Microsoft and the broader ecosystem The AI future isn’t one-size-fits-all—and neither is Azure AI Foundry. Explore Today : Azure AI Foundry
Securely Build and Manage Agents in Azure AI Foundry
Agents are transforming the way the world works, ushering in a new age of automation, efficiency, and exceptional customer experiences. These intelligent systems are revolutionizing industries, evolving from task-specific chatbots into interconnected networks of specialized agents capable of handling complex processes and adapting seamlessly to dynamic environments. But to deploy AI agents responsibly and at scale, businesses must have confidence in the underlying platform—specifically, assurance that all agent activity and customer data is secure and fully under their control. We specifically designed the new Foundry Agent Service’s standard agent setup with these requirements in mind and prioritized building new observability, evaluation, and security features. We’re also excited to introduce support for Bring Your Own (BYO) Thread Storage with Azure Cosmos DB for NoSQL as a core component of the standard agent setup. With this update, all Foundry projects created using the Standard Agent Setup will use customer managed, single tenant resources to store all customer data processed by the service. Built-in Enterprise Readiness with Foundry Agent Service Standard Setup Azure AI Foundry Agent Service offers three setup modes designed to meet you where you are—whether you're a fast-moving startup or an enterprise with strict security and compliance needs: Basic Setup: Ideal for rapid prototyping and getting started quickly, this mode uses platform-managed storage and is compatible with OpenAI Assistants. It also supports non-OpenAI models and integrates with new tools like Azure AI Search, Bing Grounding, Azure Functions, and more. Standard with Public Networking: Includes the same model and tool support as Basic Setup but gives you fine-grained control over your data by using your own Azure resources. Standard with Private Networking: Extends the Standard Setup by adding support for Bring Your Own Virtual Network (BYO VNet), enabling full network isolation and strict control over data exfiltration. Just like traditional applications, agents are stateful and require persistent storage to retain information across interactions. Azure AI Foundry Agent Service’s standard agent setup is designed for enterprise customers, and by default, requires: BYO File Storage: All files uploaded by developers (during agent configuration) or end-users (during interactions) are stored directly in the customer’s Azure Storage account. BYO Search: All vector stores created by the agent leverage the customer’s Azure AI Search resource. BYO Thread Storage: All customer messages and conversation history will be stored in the customer’s own Azure Cosmos DB account. Project-Level Data Isolation Standard setup enforces project-level data isolation by default. Two blob storage containers will automatically be provisioned in your storage account, one for files and one for intermediate system data (chunks, embeddings) and three containers will be provisioned in your Cosmos DB, one for user systems, one for system messages, and one for user inputs related to created agents such as their instructions, tools, name, etc. This default behavior was chosen to reduce setup complexity while still enforcing strict data boundaries between projects. Private Network Isolation Standard setup supports private network isolation through custom virtual network support with subnet delegation. This gives you full control over the inbound and outbound communication paths for your agent. You can restrict access to only the resources explicitly required by your agent, such as storage accounts, databases, or APIs, while blocking all other traffic by default. This approach ensures that your agent operates within a tightly scoped network boundary, reducing the risk of data leakage or unauthorized access. By default, this setup simplifies security configuration while enforcing strong isolation guarantees—ensuring that each agent deployment remains secure, compliant, and aligned with enterprise networking policies. New Foundry Resource Provider The new Foundry resource type introduces a unified management experience for agents, models, evaluations, and finetuning under a single Azure resource provider namespace. We understand the need for interconnectivity between all our offerings across AI Foundry and want to provide you with the core building blocks to use them together seamlessly. The consolidation enables administrators to apply all enterprise promises to not just agents-but all AI capabilities in your Foundry project. A few of these enterprise promises include: New built-in RBAC roles provide up-to-date role definitions to help admins differentiate access between Administrator, Project Manager and Project Users. Customer managed keys enable enterprises to bring their own encryption keys for securing sensitive agent data, ensuring compliance with internal security policies and regulatory requirements while maintaining full control over data access and lifecycle. Additionally, the new Foundry API, designed from the ground up for agentic applications, allows developers to build and evaluate across model providers using a consistent, API-first interface—further simplifying integration and accelerating development. These enhancements empower developers to accelerate experimentation and time-to-market, while giving IT admins a self-serve platform to manage agents, models, and Azure integrations cohesively. Why Should You Trust Foundry Agents? Ensuring Robust Agent Evaluation and Monitoring (AgentOps) Building trustworthy AI agents requires insight into agent decision-making processes. Azure AI Foundry Agent Services provides a comprehensive set of AgentOps tools that offer deep visibility into every stage of agent execution, enabling faster iteration, streamlined debugging, and effective evaluation. These tools include: Built-in evaluation tools allow developers to measure agent accuracy, task adherence, and overall performance under real-world conditions. This proactive approach highlights gaps and optimizes agent behavior, ensuring readiness for mission-critical tasks. Integrated OpenTelemetry-based tracing offers detailed insights into data flows, intermediate steps, and function calls during agent processes. This capability helps identify performance bottlenecks and refine workflows, ensuring seamless integration within enterprise systems. Monitoring and reporting dashboards, including Azure Monitor and Aspire, provide real-time tracking of key metrics such as response time, error rates, and task completion, enabling businesses to address issues promptly. Together, these capabilities establish a strong foundation for building secure, reliable, and scalable agentic systems. Our goal is to equip you with the tools to take your agentic applications from experimentation to production—confidently and responsibly. Strengthening Security with Entra Agent ID We are announcing the public preview of Microsoft Entra Agent ID, a new capability designed to bring enterprise-grade identity and access management to AI agents built with Azure AI Foundry and Microsoft Copilot Studio. Microsoft Entra Agent ID is the first step in managing agent identities in your organization, giving you full visibility and control over what your AI agents can do. Each agent created through Foundry Agent Service or Copilot Studio is automatically assigned a unique, first-class identity. This means your AI agents receive the same identity management as human users. They appear in your Microsoft Entra directory, allowing you to set access controls and permissions for each agent. With Entra Agent ID, organizations can: View and manage their full inventory of AI agents in one place Assign and enforce least-privilege access policies Audit agent behavior and lifecycle activity Reduce permission sprawl and limit unnecessary access Remove or restrict agents when appropriate Soon, security administrators will also be able to apply Conditional Access policies, multi-factor authentication, and role-based access controls to agents. Agent sign-in activity will be fully auditable, and agents that attempt to access unauthorized resources will be blocked just like a regular user would. Agent ID is integrated with Microsoft Defender and Microsoft Purview, enabling consistent security and compliance policies across human and non-human identities. This new capability lays the foundation for broader protection and management of digital labor as AI adoption continues to grow. Built-in Governance and Safety As organizations build agents robust safety and governance controls are essential to ensuring responsible AI behavior. Microsoft is announcing several new capabilities designed to help teams address emerging risks such as prompt injection attacks, privacy violations, and agent drift from intended tasks. These guardrails and controls are seamlessly integrated into the agent service, powered by Azure AI Content Safety. At Build, we are introducing three critical advancements: Spotlighting in Prompt Shields: Enhances the ability to separate intended user instructions from potentially malicious or untrusted content, such as information pulled from documents or websites. This separation helps reduce the risk of cross prompt injection attacks. PII Detection: Adds new data protection capabilities, powered by Azure AI Language, that automatically detect and redact PII, PHI, and other sensitive information from unstructured text. This helps safeguard privacy and reduce the risk of data exposure in AI outputs. Task Adherence: Control to detect when an agent strays from user intent. Deviations can be blocked or escalated, helping agents follow instructions and stay within approved boundaries. Conclusion The future of AI depends on trust and collaboration—only with both can scalable systems truly redefine workflows and unlock groundbreaking solutions. Azure AI Foundry is empowering organizations to step boldly into this future, unlocking the limitless possibilities of AI agents to shape a smarter, more connected world. Whether you're deploying an agent to deliver personalized shopping recommendations or to process confidential legal documents, each use case requires a different level of security, access control, and system safeguards. That’s why we’ve built transparency and control into the foundation of our platform—so you can tailor your deployment to match your specific risk profile and operational needs. Get started today by deploying one of our one-click “Deploy to Azure” ARM templates. What’s Next? Build your first network secured Agent through ARM template Explore the documentation to learn more about Azure AI Foundry Agent Service Start building your agents today in Azure AI Foundry Watch our Foundry Agent Service breakout session at BuildThe Future of AI: How Lovable.dev and Azure OpenAI Accelerate Apps that Change Lives
Discover how Charles Elwood, a Microsoft AI MVP and TEDx Speaker, leverages Lovable.dev and Azure OpenAI to create impactful AI solutions. From automating expense reports to restoring voices, translating gestures to speech, and visualizing public health data, Charles's innovations are transforming lives and democratizing technology. Follow his journey to learn more about AI for good.The Future of AI: Autonomous Agents for Identifying the Root Cause of Cloud Service Incidents
Discover how Microsoft is transforming cloud service incident management with autonomous AI agents. Learn how AI-enhanced troubleshooting guides and agentic workflows are reducing downtime and empowering on-call engineers.Introducing Phi-4: Microsoft’s Newest Small Language Model Specializing in Complex Reasoning
Today we are introducing Phi-4, our 14B parameter state-of-the-art small language model (SLM) that excels at complex reasoning in areas such as math, in addition to conventional language processing. Phi-4 is the latest member of our Phi family of small language models and demonstrates what’s possible as we continue to probe the boundaries of SLMs. Phi-4 is available on Azure AI Foundry and on Hugging Face. Phi-4 Benchmarks Phi-4 outperforms comparable and larger models on math related reasoning due to advancements throughout the processes, including the use of high-quality synthetic datasets, curation of high-quality organic data, and post-training innovations. Phi-4 continues to push the frontier of size vs quality. Phi-4 is particularly good at math problems, for example here are the benchmarks for Phi-4 on math competition problems: Phi-4 performance on math competition problems To see more benchmarks read the newest technical paper released on arxiv. Enabling AI innovation safely and responsibly Building AI solutions responsibly is at the core of AI development at Microsoft. We have made our robust responsible AI capabilities available to customers building with Phi models, including Phi-3.5-mini optimized for Windows Copilot+ PCs. Azure AI Foundry provides users with a robust set of capabilities to help organizations measure, mitigate, and manage AI risks across the AI development lifecycle for traditional machine learning and generative AI applications. Azure AI evaluations in AI Foundry enable developers to iteratively assess the quality and safety of models and applications using built-in and custom metrics to inform mitigations. Additionally, Phi users can use Azure AI Content Safety features such as prompt shields, protected material detection, and groundedness detection. These capabilities can be leveraged as content filters with any language model included in our model catalog and developers can integrate these capabilities into their application easily through a single API. Once in production, developers can monitor their application for quality and safety, adversarial prompt attacks, and data integrity, making timely interventions with the help of real-time alerts. Phi-4 in action One example of the mathematical reasoning Phi-4 is capable of is demonstrated in this problem. Start Exploring Phi-4 is currently available on Azure AI Foundry and Hugging Face, take a look today.
Introducing Evaluation API on Azure OpenAI Service
We are excited to announce new Evaluations (Evals) API in Azure OpenAI Service! Evaluation API lets users test and improve model outputs directly through API calls, making the experience simple and customizable for developers to programmatically assess model quality and performance in their development workflows.
