Broad model catalog and model customization
Access foundational, open, reasoning, multimodal, and industry-specific models from providers listed by Microsoft, then train, fine-tune, distill, or upgrade models with minimal coding.
Microsoft Foundry
Freemium
Microsoft Foundry is an enterprise AI platform for building, grounding, deploying, and governing AI applications and agents. It helps development and data teams connect models, organizational knowledge, business tools, and lifecycle controls in Azure.
什么是 Microsoft Foundry？
Microsoft Foundry is an enterprise AI platform for building, grounding, deploying, and governing AI applications and agents. It combines model access and customization, agent development, organizational knowledge, business tools, observability, and trust controls in a unified Azure environment.
The platform is designed for development and data teams creating AI-driven, cloud-native applications. Teams can choose from a broad model catalog, build agents with Foundry Agent Service or the open-source Microsoft Agent Framework, connect agents to organizational and business-system context, and manage production behavior through monitoring and governance capabilities.
Foundry pricing is consumption-based across the services and features used. Each service has its own billing model and price, so the total cost depends on the selected capabilities and usage.
Microsoft Foundry 能做什么？
Real-time model routing
Use intelligent model routing to select a model for a particular task and optimize performance or cost without rewriting the application.
Agent development and hosting
Build action-oriented, context-aware agents with Foundry Agent Service and use the open-source Microsoft Agent Framework to create, orchestrate, and extend agent applications. Foundry also supports serverless containers and event-driven functions for hosting and scaling.
Grounding and business-system connectivity
Ground agents in organizational intelligence with permission-aware grounding and Microsoft Graph integration. Connect workflows to business systems through more than 1,400 prebuilt connections, including SAP, Salesforce, and Dynamics 365, or add custom tools through Model Context Protocol.
Observability and evaluation
Trace agent interactions in production with OpenTelemetry-based tooling and evaluate applications using built-in measures for coherence, relevance, groundedness, and safety.
Unified trust and governance controls
Manage agents, tools, and knowledge sources through a control plane integrated with Microsoft Entra, Purview, and Defender. Runtime controls can help simulate, detect, and mitigate prompt attacks, hallucinations, and sensitive-data leakage.
使用场景
“Automate complex business processes”
Create agents that perform specific, action-oriented tasks across connected business systems while retaining human control over the workflow.
“Build grounded enterprise assistants”
Combine permission-aware organizational context, Microsoft Graph, and business data so applications can reason using relevant organizational signals rather than relying only on a general model.
“Develop and modernize AI applications”
Use model choice, routing, agent frameworks, prebuilt tools, and Azure services to add AI capabilities to new or existing cloud-native applications.
“Operate AI agents in production”
Trace interactions, evaluate response quality and safety, and apply runtime trust controls as agent applications move from development into ongoing use.
“Create data and document workflows”
Combine customizable tools such as OCR, translation, speech, and object detection with agents or applications to automate information-processing workflows.
常见问题
What is Microsoft Foundry used for?
It is used to build, ground, deploy, monitor, and govern AI applications and agents. Common workflows include business-process automation, data and document processing, application modernization, and adding AI capabilities to cloud-native applications.
Can Microsoft Foundry use models from different providers?
Yes. Microsoft describes Foundry Models as a catalog spanning foundational, open, reasoning, multimodal, and industry-specific models, including models from providers such as OpenAI, Anthropic, Meta, Google, xAI, and Hugging Face.
How can agents access business context and tools?
Agents can use permission-aware grounding, Microsoft Graph integration, Microsoft IQ, Foundry IQ, and Fabric IQ. They can also connect to business systems through prebuilt connections, use customizable tools, or access custom APIs through Model Context Protocol.
How does Microsoft Foundry support production oversight?
Its control-plane capabilities include OpenTelemetry-based tracing, evaluators for coherence, relevance, groundedness, and safety, and runtime controls intended to help detect and mitigate prompt attacks, hallucinations, and sensitive-data leakage.
How is Microsoft Foundry priced?
Pricing is consumption-based and depends on the individual Foundry services and features used. Each service has its own billing model and price; Microsoft provides pricing estimates, a pricing calculator, and sales support for quotes.
快速信息
- Category
- Enterprise AI platform
- Platform
- Microsoft Azure
- Primary users
- Developers, data scientists, and organizations building AI applications and agents
- Development options
- Foundry Agent Service, Microsoft Agent Framework, prebuilt tools, and custom MCP tools
- Governance
- Unified control plane with Microsoft Entra, Purview, and Defender integrations
- Pricing model
- Consumption-based; individual services and features have separate billing models
Microsoft Foundry 替代品
IBM watsonx.ai
www.ibm.com
IBM watsonx.ai is an enterprise AI development studio for building predictive, prescriptive, and generative AI solutions. It supports AI builders, data scientists, and developers across model development, customization, retrieval-augmented generation, deployment, and lifecycle management.
Gemini Enterprise Agent Platform
cloud.google.com
Gemini Enterprise Agent Platform, formerly Vertex AI, is Google Cloud’s platform for developers and technical teams to build, deploy, govern, and optimize AI agents, generative AI applications, and machine learning models.
Together AI
together.ai
Together AI 是一个支持推理、微调、GPU 集群、沙盒和托管存储的 AI 云平台。
Amazon Bedrock
aws.amazon.com
Amazon Bedrock is a fully managed AWS platform for building generative AI applications and agents with access to foundation models, customization tools, safety controls, and production-oriented workflows. It supports teams that want to experiment through the console or build applications through AWS APIs and SDKs.
Bento
www.bentoml.com
Bento is an inference platform for packaging, deploying, optimizing, and operating AI and machine-learning models at scale. It supports open and custom models across cloud, on-premises, Kubernetes, and bring-your-own-cloud environments.
Prodia
prodia.com
Prodia is a multi-silicon inference platform focused on video generation. It develops AI model implementations across different hardware to balance cost, output quality, and performance.