GPT-5 in Azure AI Foundry: Built for real-world workloads<\/h2>\n\n\n\n
In Azure AI Foundry, the GPT-5 models are available via API and orchestrated by themodel router. The GPT-5 series spans complementary strengths:<\/p>\n\n\n\n
- \n
- GPT-5<\/strong>, a full reasoning model provides deep, richer reasoning for analytics and complex tasks, like code generation, with a 272k token context.<\/li>\n\n\n\n
- GPT-5 mini<\/strong> powers real-time experiences for apps and agents that require reasoning, tool calling to solve customer problems.<\/li>\n\n\n\n
- GPT-5 nano<\/strong> is a new class of reasoning model which focuses on ultra-low-latency and speed with rich Q&A capabilities.<\/li>\n\n\n\n
- GPT-5 chat<\/strong> enables natural, multimodal, multi-turn conversations that remain context-aware throughout agentic workflows, with 128k token context.<\/li>\n<\/ul>\n\n\n\n
Together, the suite delivers a seamless continuum from rigorous agentic coding tasks, to relatively simple Q&A\u2014all delivered with the same Azure AI Foundry endpoint using model router in Foundry Models.<\/p>\n\n\n\n
Under the hood, GPT-5 unifies advanced reasoning, code generation, and natural language interaction. It combines analytical depth with intuitive dialogue to solve end-to-end problems and explain its approach. Agentic capabilities allow multi-step tool use and long action chains with transparent, auditable decisions. As a frontier-level coding model, GPT-5 can plan complex agentic workflow, build migrations, and refactor code, as well as produce tests and documentation with clear rationale. Developer controls\u2014including parameters like
reasoning_effort<\/code> <\/strong>and verbosity\u2014let teams tune depth, speed, and detail, while new freeform tool-calling features<\/strong> enable broadens tool compatibility without rigid schemas.<\/p>\n\n\n\nOrchestrate with the model router\u2014then scale with agents<\/h2>\n\n\n\n
Introducing GPT-5 to Azure AI Foundry is more than a model drop: it\u2019s a leap forward for the platform. Starting today, developers can use the model router in Foundry Models to maximize the capabilities of the GPT-5 family models (and other models in Foundry Models) while saving up to 60% on inferencing cost with no loss in fidelity. Powered by a fine-tuned SLM under the hood, the model router evaluates each prompt and decides the optimal model based on the complexity, performance needs, and cost efficiency of each task. Let the model router pick the right model so that you can build your AI-powered applications with ease.<\/p>\n\n\n\n
\nLearn about GPT-5 in Azure AI Foundry featuring model router<\/a><\/div>\n<\/div>\n\n\n\nAnd orchestration doesn\u2019t stop at routing\u2014Foundry carries the same intelligence into agents. Coming soon, GPT-5 will be available in the Foundry Agent Service, pairing frontier models with built-in tools including new browser automation<\/a> and Model Context Protocol (MCP) integrations. The result: policy-governed, tool-using agents that can search, act in web apps, and complete end-to-end tasks\u2014instrumented with Foundry telemetry and aligned to Microsoft Responsible AI.<\/p>\n\n\n\n
Accelerating business impact with GPT-5<\/h2>\n\n\n\n
These capabilities map directly to business impact.<\/p>\n\n\n\n
In research and knowledge work, GPT-5 accelerates financial and legal analysis, market intelligence, and due diligence\u2014reading at scale and producing decision-ready output with traceability. In operations and decisioning, it strengthens logistics support, risk assessment, and claims processing by pairing robust reasoning with policy adherence. Copilots and customer experience teams benefit from multi-turn, multimodal agents that reason in real time, call tools, resolve tasks, and revert to humans with more helpful context.<\/p>\n\n\n\n
In software engineering, GPT-5 excels at code generation, application modernization, and quality engineering\u2014improving code style and explanations to compress review cycles.<\/p>\n\n\n\n
And for use cases which are cost or latency sensitive, GPT-5-nano\u2019s ultra\u2011low\u2011latency architecture delivers rapid, high\u2011accuracy responses, making it the ideal target for fine\u2011tuning and the go\u2011to model for high\u2011volume, straightforward requests.<\/p>\n\n\n\n
GPT-5 customer spotlight <\/h2>\n\n\n\n
Customers are unleashing GPT-5 across complex, mission-critical workloads\u2014accelerating decision-making, supercharging coding, and catalyzing product innovation.<\/p>\n\n\n\n
SAP<\/h3>\n\n\n\n
\n
SAP is excited to be among the first to leverage the power of GPT-5 in Azure AI Foundry within our generative AI hub in AI Foundation. GPT-5 in Azure AI Foundry will enable our product team and our developer community to deliver impactful business innovations to our customers.<\/em><\/p>\n\u2014Dr. Walter Sun, SVP and Global Head of AI, SAP SE <\/cite><\/blockquote>\n\n\n\n
Relativity<\/h3>\n\n\n\n
\n
The GPT-5 in Azure AI Foundry raises the bar for putting legal data intelligence into action\u2026 This next-generation AI will empower legal teams to uncover deeper insights, accelerate decision-making, and drive stronger strategies across the entire legal process.<\/em><\/p>\n\u2014Dr. Aron Ahmadia, Senior Director, Applied Science, Relativity <\/cite><\/blockquote>\n\n\n\n
Hebbia<\/h3>\n\n\n\n
\n
The partnership between Hebbia and Azure AI Foundry gives financial professionals an unprecedented edge. With GPT-5’s advanced reasoning in Hebbia, they can pinpoint critical figures across thousands of documents and structuring complex financial analysis with speed and accuracy.<\/em><\/p>\n\u2014Danny Wheller, VP of Business and Strategy<\/cite><\/blockquote>\n\n\n\n
Building with AI in GitHub Copilot and Visual Studio Code<\/h2>\n\n\n\n
GPT-5 begins rolling out today to millions of developers using GitHub Copilot<\/a> <\/strong>and Visual Studio Code<\/a><\/strong>, applying the flagship model\u2019s advanced reasoning capabilities to increasingly complex problems\u2014from sophisticated refactoring to navigating large codebases more effectively. GPT-5 helps developers write, test, and deploy code faster, while supporting agentic coding tasks with significant improvements to coding style and overall code quality. With GPT-5, developers not only code faster, but code better.<\/p>\n\n\n\n
![\"A](\"https:\/\/azure.microsoft.com\/en-us\/blog\/wp-content\/uploads\/2025\/08\/VSCode_GPT-5_1920x1080VidSocialPost-GIF.gif\")
<\/figure>\n\n\n\nWith today\u2019s VS Code release, developers also gain a more powerful agentic coding experience directly within the editor: GitHub Copilot\u2019s coding agent has an improved experience for autonomously tackling tasks in the background. Additionally, the GitHub Copilot chat experience brings increased productivity, including support beyond 128 tools for a single chat request and chat checkpoints allowing users to restore workspace changes to a prior point. Today, we are also announcing an updated extension to develop agents using the Azure AI Foundry extension all within VS Code environment.<\/p>\n\n\n\n
These announcements extend Microsoft’s strategy to transform software development with AI, bringing advanced AI capabilities to the entire software lifecycle.<\/p>\n\n\n\n
Security, safety, and governance by design<\/h2>\n\n\n\n
In all domains, security and safety is a layer cake of protections, which together provide protection for risk scenarios\u2014and AI is no different. For AI, we think about layers with the model as the core. With GPT-5, the core is safer than before:<\/p>\n\n\n\n
\n
The Microsoft AI Red Team found GPT-5 to have one of the strongest safety profiles of any OpenAI model, performing on par with\u2014or better than\u2014o3.<\/em><\/p>\n\u2014Dr. Sarah Bird, Chief Product Officer of Responsible AI, Microsoft<\/cite><\/blockquote>\n\n\n\n
As we think about the safety, security, and governance layers around this core\u2014Azure AI Foundry <\/strong>provides a number of additional controls:<\/p>\n\n\n\n
- \n
- Azure AI Content Safety<\/strong> protections are applied to every prompt and completion, such as prompt shields, which help to detect and mitigate prompt-injection attempts before they reach the model.<\/li>\n<\/ul>\n\n\n\n
![\"A](\"https:\/\/azure.microsoft.com\/en-us\/blog\/wp-content\/uploads\/2025\/08\/foundry-image.jpg\")
<\/figure>\n- \n
- Built-in agent evaluators work with the AI Red Teaming Agent<\/strong> to run alignment, bias, and security tests throughout development and production, while continuous evaluation <\/strong>streams real-time metrics\u2014latency, quality, safety, and fairness\u2014stream into Azure Monitor and Application Insights for single-pane visibility.<\/li>\n<\/ul>\n<\/div><\/div>\n\n\n\n
- \n
- Finally, security signals integrate directly with Microsoft Defender for Cloud<\/strong>, and runtime metadata and evaluation results are integrated to Microsoft Purview<\/strong> for audit, data-loss prevention, and regulatory reporting, extending protection and governance across the entire GPT-5 lifecycle.<\/li>\n<\/ul>\n\n\n\n
Bringing AI into every workflow with GitHub Copilot and Visual Studio Code<\/h2>\n\n\n\n
Starting today, GPT-5 begins rolling out to millions of developers<\/strong> who use GitHub Copilot<\/strong> and Visual Studio Code who will be able to select GPT-5 to write, test, and deploy code<\/strong>\u2014and develop agents using the Azure AI Foundry extension all within VS Code environment. GPT-5 supports complex agentic coding tasks with significant improvements to coding personality, front-end aesthetics, and code quality, highly desired improvements for the developer community.<\/p>\n\n\n\n
Our evaluations show OpenAI GPT-5\u2019s reasoning capabilities and contextual awareness exceed o3, enabling developers to tackle more complex problems\u2014from refactoring to navigating large codebases. With GPT-5, users in the Visual Studio family can not only code faster, but code better.<\/p>\n\n\n\n
VS Code and our recent decision to open-source GitHub Copilot<\/strong><\/a>, represents our commitment to open tools and standards and demonstrates our ability to meet the rapid pace of model innovations while keeping the developer experience at the forefront. In today\u2019s release of VS Code, developers can. In today\u2019s VS Code release<\/strong><\/a>, developers have even more control over their experience in chat\u2014with improvements to the reliability of terminal tools, updates to the tool picker and limits, new checkpoints, and more.<\/p>\n\n\n\n
Today’s announcement extends Microsoft’s strategy to transform software development with AI, bringing advanced AI capabilities to the entire software lifecycle.<\/p>\n\n\n\n
Start building today<\/h2>\n\n\n\n
GPT-5 is available via our Standard offering in Azure AI Foundry, with deployment choices optimized for cost-efficiency and governance needs, including Global and Data Zone (United States, European Union) deployment options for data residency and compliance.1<\/sup><\/p>\n\n\n
![\"A](\"https:\/\/azure.microsoft.com\/en-us\/blog\/wp-content\/uploads\/2025\/08\/Screenshot-2025-08-07-152630.webp\")
<\/figure>\n\n\n\nWith Azure AI Foundry\u2019s first-class reliability, realtime evaluations, built-in observability, and secure deployment options, you can confidently move from pilot to production\u2014all aided while unique tools like Model Router optimizes quality, latency, and cost across workloads.<\/p>\n\n\n\n
- Finally, security signals integrate directly with Microsoft Defender for Cloud<\/strong>, and runtime metadata and evaluation results are integrated to Microsoft Purview<\/strong> for audit, data-loss prevention, and regulatory reporting, extending protection and governance across the entire GPT-5 lifecycle.<\/li>\n<\/ul>\n\n\n\n
- Built-in agent evaluators work with the AI Red Teaming Agent<\/strong> to run alignment, bias, and security tests throughout development and production, while continuous evaluation <\/strong>streams real-time metrics\u2014latency, quality, safety, and fairness\u2014stream into Azure Monitor and Application Insights for single-pane visibility.<\/li>\n<\/ul>\n<\/div><\/div>\n\n\n\n
- GPT-5 mini<\/strong> powers real-time experiences for apps and agents that require reasoning, tool calling to solve customer problems.<\/li>\n\n\n\n