Microsoft have announced general availability of OpenAI's GPT-5 in Azure AI Foundry, described as "the most powerful LLM ever released across key benchmarks" with frontier reasoning capabilities and enterprise-grade platform integration.
The GPT-5 series includes four complementary models optimised for different workloads. GPT-5 provides deep reasoning for analytics and complex tasks like code generation with 272k token context. GPT-5 mini powers real-time experiences for apps and agents requiring reasoning and tool calling. GPT-5 nano represents "a new class of reasoning model" focusing on ultra-low-latency and speed with rich Q&A capabilities. GPT-5 chat enables natural, multimodal, multi-turn conversations with 128k token context for agentic workflows.
The platform introduces a model router powered by a fine-tuned SLM that evaluates prompts and selects optimal models based on complexity, performance needs, and cost efficiency. This routing system delivers up to 60% savings on inferencing costs with no loss in fidelity across the GPT-5 family and other Foundry Models.
GPT-5 will be available in the upcoming Foundry Agent Service, featuring built-in tools including browser automation and Model Context Protocol integrations. The service provides policy-governed, tool-using agents capable of searching, acting in web apps, and completing end-to-end tasks with Foundry telemetry and Microsoft Responsible AI alignment.
Customer implementations span multiple sectors. SAP leverages GPT-5 within their generative AI hub in AI Foundation for business innovations. Relativity applies the model for legal data intelligence to accelerate decision-making and drive strategies across legal processes. Hebbia uses GPT-5's advanced reasoning for financial analysis, enabling professionals to pinpoint critical figures across thousands of documents.
GPT-5 begins rolling out to millions of developers using GitHub Copilot and Visual Studio Code, applying advanced reasoning to complex problems including sophisticated refactoring and large codebase navigation.
The platform targets research and knowledge work for financial and legal analysis, operations and decisioning for logistics support and risk assessment, customer experience teams with multi-turn multimodal agents, and software engineering for code generation and application modernisation. Cost-sensitive use cases benefit from GPT-5-nano's ultra-low-latency architecture.
Azure AI Foundry's model router technology enables organisations to optimise quality, latency, and cost across workloads while maintaining enterprise security through Azure AI Content Safety protections, built-in agent evaluators, and integration with Microsoft Defender for Cloud and Microsoft Purview for audit and regulatory reporting.