Azure AI Foundry's GPT-5: The future of AI apps and agents begins here

Today we are announcing the general availability of Openai's new flagship, GPT-5, at Azure AI Foundry. This is more than a new model release. This is the most powerful major language model (LLM) ever released across key benchmarks.

For business leaders building with AI, conversations go beyond chat. The bar is high. Can AI generate, infer and provide measurable results?

Today we announce general availability Openai's new flagship model for Azure AI Foundry, GPT-5. This is more than a new model release. This is the most powerful LLM ever released across the key benchmark. Azure AI Foundry's GPT-5 combines high-performance generation and cost-effective frontier inference offered by Microsoft Azure's enterprise-grade platform to enable organizations to move confidently from pilots to production.

Azure AI Foundry GPT-5: Built for real workloads

In Azure AI Foundry, the GPT-5 model is available via API and is tuned by the Themodel router. The GPT-5 series ranges in complementary strength.

GPT-5,The complete inference model provides deeper and richer inference for analysis and complex tasks such as code generation with a 272k token context.
GPT-5 Mini Enhance the real-time experience for apps and agents that require inference and tool calls to resolve customer problems.
GPT-5 Nano A new class of inference models focusing on ultra-low latency and speed with a wealth of Q&A features.
GPT-5 Chat It allows natural, multimodal, multi-turn conversations that keep context aware throughout the agent workflow with a 128K token context.

Together, the suite offers a seamless continuum from strict agent coding tasks to relatively simple Q&A.

Under the hood, GPT-5 unifies advanced inference, code generation, and natural language interactions. Combining the depth of analysis with intuitive dialogue, it solves end-to-end problems and explains its approach. The agent functionality allows for a long action chain with the use of multi-step tools and transparent, auditable decisions. As a frontier-level coding model, GPT-5 can plan complex agent workflows, build migrations, refactor code, and create tests and documentation with clear evidence. Developer Control – Includes parameters like reasoning_effort And redundancy – teams adjust depth, speed and details New Freeform Tool Call Function Allows increased tool compatibility without a rigid schema.

Adjust with the model router. Then use the agent

Deploying GPT-5 with Azure AI Foundry is more than a model drop. It's a jump for the platform. Starting today, developers can use model routers with casting models to save up to 60% in guess costs without compromising fidelity while maximizing the functionality of the GPT-5 family models (and other models in the foundry models). A model router with fine-tuned SLM under the hood evaluates each prompt and determines the best model based on the complexity of each task, performance needs and cost-effectiveness. Make your model routers easy to choose the right model and build applications powered by AI.

Orchestration does not stop with routing. Foundry brings the same intelligence to the agent. The upcoming GPT-5 will be available in the Foundry Agent Service, combining frontier models with built-in tools such as new browser automation and Model Context Protocol (MCP) integration. The result: Agents that dominate tool use agents installed with foundry telemetry and aligned with Microsoft Responsibility AI, search, web apps, and fully search for end-to-end tasks.

Accelerating the impact of GPT-5 on business

These features directly map business impacts.

In the research and knowledge work, GPT-5 accelerates financial and legal analysis, market information, and due diligence. It produces large-scale reads and traceable decision-making outcomes. Enhance logistics support, risk assessment, and claims handling by combining robust inference and policy compliance in operations and decision-making. Copilots and Customer Experience teams benefit from multi-turn multimodal agents inference in real time, invoking tools, solving tasks, and returning them to humans in a more beneficial context.

In software engineering, GPT-5 is excellent at code generation, application modernization, and quality engineering. This improves the code style and description for compressing the review cycle.

Additionally, for cost- or latency-sensitive use cases, the GPT-5-Nano's ultra-low latency architecture provides quick, accurate response, an ideal target for fine-tuning and a Go-to model for large amounts of simple requests.

GPT-5 Customer Spotlight

Customers unleash GPT-5s across complex, mission-critical workloads. Catalysts for decision-making access, supercharged coding, and product innovation.

Sap

SAP is excited to first leverage the power of Azure AI Foundry's GPT-5 within the AI Foundation Generation AI Hub. Azure AI Foundry's GPT-5 enables product teams and developer community to deliver impactful business innovation to our customers.

– dr. Walter Sun, SVP and Global Head of AI, SAP SE

Relativity Theory

Azure AI Foundry's GPT-5 procures the criteria for implementing legal data intelligence. This next-generation AI will enable legal teams to uncover deeper insights, accelerate decision-making, and drive stronger strategies throughout the legal process.

– dr. Senior Director of Aron Ahmadia, Applied Science and Relativity

Heavy

The partnership between Hebia and Azure AI Foundry gives financial experts an unprecedented advantage. Advanced inference of GPT-5 in Hebbia allows them to identify important figures across thousands of documents and build complex financial analyses with speed and accuracy.

– Danny Waller, Vice President of Business and Strategy

Buildings using AI in Github Copilot and Visual Studio Code

GPT-5 is starting to roll out to millions of developers today Github Copilot and Visual Studio Codefrom sophisticated refactoring to more effective navigating large codebases, it applies the advanced inference capabilities of flagship models to increasingly complex problems. GPT-5 helps developers write, test and deploy code faster, supporting agent coding tasks that have significantly improved coding style and overall code quality. With GPT-5, developers not only code faster, but also write better code.

With today's code release, developers will get a more powerful agent coding experience directly within the editor. GithubCopilot's coding agents provide an improved experience for working on tasks autonomously in the background. Additionally, GitHub Copilot Chat Experience brings productivity gains, including support for over 128 tools for a single chat request and chat checkpoint support, allowing users to restore workspace changes to their previous points. Today we are also announcing updated extensions for developing agents all within VS code environments using the Azure AI Foundry extension.

These announcements will extend Microsoft's strategy to transform software development with AI, bringing advanced AI capabilities to the entire software lifecycle.

Designed security, safety and governance

In all domains, security and safety are layer cakes of protection, providing together protection for risk scenarios, AI is no exception. In the case of AI, we consider the model's layers to be the core. With the GPT-5, the core is safer than before.

The Microsoft AI Red team has discovered that the GPT-5 has one of the most powerful safety profiles of the Openai model.

– dr. Sarah Bird, Microsoft's Chief Product Officer of Responsible AI

Considering the safety, security and governance layers around this core,Azure AI Foundry It provides many additional controls:

Azure AI content safety Protection applies to all rapid completions, including prompt shields, which help to detect and mitigate prompt injection attempts before reaching the model.

Embedded Agent Evaluator works with AI Red Team Agent To carry out alignment, bias and security testing throughout development and production, Continuous evaluation Stream real-time metrics (latency, quality, safety, fairness) and flow into Azure monitor and application insights for single-panel visibility.

Finally, the security signals are directly integrated Microsoft Defender in Cloudand runtime metadata and evaluation results are combined into Microsoft Purview Extend protection and governance throughout the GPT-5 lifecycle for auditing, data loss prevention and regulatory reporting.

Bring AI into all your workflows using Github Copilot and Visual Studio code

Starting today, GPT-5 will begin to roll out. Millions of developers Who will use it? Github Copilot And you can visual studio codes Select GPT-5 to write, test, and expand the code– and use Azure AI Foundry extensions to develop agents all within a VS code environment. GPT-5 supports complex agent coding tasks that have been significantly improved to coding personality, front-end aesthetics, code quality, and improvements that are highly desirable for the developer community.

Our evaluation shows OpenAI GPT-5's inference ability and context awareness exceed O3, allowing developers to tackle more complex issues. GPT-5 allows users of the Visual Studio family to code faster as well as better code.

VS Code and Us Recent decisions for open source Github Copilotrepresenting our commitment to opening tools and standards, demonstrating our ability to meet the rapid pace of model innovation while keeping the developer experience at the forefront. Today's VS Code release allows developers to do it. in Today's code releasedevelopers can further control their chat experience by improving terminal tools reliability, tool picker updates and restrictions, new checkpoints and more.

With today's announcement, Microsoft's strategy will transform software development with AI, bringing advanced AI capabilities to the entire software lifecycle.

Start building today

GPT-5 is available through our standard products in Azure AI Foundry, and deployment options are optimized for cost-effectiveness and governance needs, including global and data zone (US, European Union) deployment options for data residency and compliance.¹

With Azure AI Foundry's top-notch reliability, real-time assessment, built-in observability, and secure deployment options, unique tools such as model routers optimize quality, latency and cost across your workload.