The best model for your business will keep changing. Adopting it should move your business forward, not send your team back to rebuild the architecture around it. As models advance, organizations need the freedom to choose the right model for each workload, and an agent foundation that can evolve with those choices.
Microsoft Foundry provides that model—and harness—agnostic foundation for building and running agents. Teams can adopt better models as they emerge while preserving their investments in the enterprise systems, knowledge, tools, and controls that make agents useful to their business. Adopting a newer model is only the starting point.
Foundry helps teams continuously improve their agents, using production traces to evaluate and refine instructions, skills, tools, and model choice against quality, latency, and cost. That’s the hill-climbing approach: observe, evaluate, optimize, validate, and repeat—with people in control Learn more about Microsoft Foundry Today, we’re advancing that foundation to help teams build and continuously improve agents that deliver business value: Choose from a broader range of models as frontier AI advances.
Build voice experiences and support long-running work on the same agent foundation. Equip agents with knowledge and tools that help them discover and apply the right capabilities. Turn production insights into continuous improvement across agent quality, latency, and cost.
These advances help teams spend less time assembling infrastructure and more time building. One customer putting these capabilities together is Fashable , which uses Foundry’s models, agents, knowledge, and tools to help fashion brands turn trends and ideas into clothing designs, virtual try-on experiences, and campaign imagery. By connecting these workflows with the existing brands’ systems, Fashable reports cutting the time to develop new fashion products from months to weeks, with one customer reducing its budget for physical samples by 60%.
Choose and evaluate frontier AI models in Microsoft Foundry Frontier AI is advancing rapidly across providers. New capabilities are arriving in days, not months, and this week captures that momentum: the full GPT-6 family from OpenAI , including GPT-6 Sol and GPT-6 Luna, and Claude Opus 5. 5 from Anthropic are now available in Microsoft Foundry.
What makes this moment important is not any single model. It is the pace and breadth of innovation happening across the frontier at the same time. For developers and enterprises, model choice is no longer a decision you make at the start of a project.
It is becoming a continuous advantage. The best model for a coding agent today may not be the best model for research, customer service, or high-volume inference, and the right answer will keep changing as new models arrive. Microsoft Foundry is built for that reality.
Foundry brings together models from leading AI providers in an open platform where teams can evaluate them against their own data and workloads, compare tradeoffs across quality, latency, and cost, and put the right models into production without rebuilding their AI strategy around a single provider. The arrival of the GPT-6 family and Claude Opus 5. 5 expands that choice, giving teams more models to evaluate for their specific workloads alongside the broader portfolio already available in Foundry.
Build and operate voice agents with Foundry Agent Service Voice is rapidly becoming a major way people interact with AI agents. Yet many voice experiences are still built by layering speech onto agents designed for text, requiring separate speech, orchestration, monitoring, and deployment components. With voice agents in Foundry Agent Service , voice becomes a native part of the agent foundation—built, deployed, observed, evaluated, and optimized through the same platform, API, and SDK as the rest of your Foundry agents.
Voice agents, now in public preview, are a new agent type supporting both prompt and hosted agents. Choose from industry-leading voice models— GPT Realtime, Azure Realtime, MAI and more—to balance latency, reasoning depth, and cost for each scenario, or bring your own. Models, tools, and deployment paths are designed for spoken interaction from the start: expressive voices, natural turn-taking, and graceful recovery when someone interrupts, across 80+ languages and 140+ locales.
Teams that want a distinct identity can fine-tune custom speech for domain vocabulary, create a custom voice, or add a photo or video avatar for unique brand identity. Developers can reuse the tools, enterprise knowledge, and governance configured for their other Foundry agents, then deploy to web channels, Microsoft Teams, Teams Phone, and Twilio-based inbound and outbound telephony experiences.
The developer workflow extends across the tools teams already use: AZD AI provides a repeatable, source-controlled workflow for configuring and deploying voice agents. Foundry Toolkit in Visual Studio Code ( coming soon ) enables developers to create, run, speak with, and debug voice agents without leaving their development environment. Voice-agent observability brings tracing, monitoring, transcripts, and voice-native evaluation to spoken interactions.
At Fashable, our mission is to make advanced AI accessible to brands and manufacturers of all sizes, regardless of their technical maturity. As we introduce new agentic capabilities, voice becomes an important part of that journey, enabling interactions that feel more natural, conversational, and accessible. People don’t think in prompts, they think in conversations, and using voice agents in Foundry with MAI-Voice helps bridge that gap.
—Orlando Ribas Fernandes, CEO, Fashable Build resilient hosted agents for long-running tasks Real enterprise work does not always finish within a single request. A research task may span many sources, a report may require several tools and processing stages, and an agent may need to wait for external systems or human approval.
Originally published at azure.microsoft.com
