Foundry Models
Model catalog (11,000+ models from Azure, Anthropic, Meta, Mistral, Hugging Face and others) with managed-compute and serverless per-token deployment.
Find alternatives to Foundry Models
Own page's 'Billing for serverless deployments' section, live 2026-09-14: serverless models are billed per-token ("billed for inputs and outputs to the APIs, typically in tokens"), managed compute is billed by "virtual machine core hours", and "Foundry Models sold by Azure are billed via Azure meters as First Party Consumption Services" while partner/community models are "billed through Azure Marketplace". Distinct mechanism from Content Safety's per-unit metering — not shared text. Recurring free access is not stated either way.
What it does
- Model inference
- Model hosting
Sources
- Pricing
-
Own page's 'Billing for serverless deployments' section, live 2026-09-14: serverless models are billed per-token ("billed for inputs and outputs to the APIs, typically in tokens"), managed compute is billed by "virtual machine core hours", and "Foundry Models sold by Azure are billed via Azure meters as First Party Consumption Services" while partner/community models are "billed through Azure Marketplace". Distinct mechanism from Content Safety's per-unit metering — not shared text.
- Description
-
Official documentation · 23 Sep 2026
Foundry Models' own product page: "Innovate faster using more than 11,000+ models packed for out-of-the-box-use"; serverless and managed-compute deployment. The old learn.microsoft.com url now lands on the '(classic)' portal docs.
- What it does
-
Official documentation · 23 Sep 2026
Foundry Models' own product page: "Innovate faster using more than 11,000+ models packed for out-of-the-box-use"; serverless and managed-compute deployment. The old learn.microsoft.com url now lands on the '(classic)' portal docs.
- Sold within
-
Own page, live 2026-09-14: "Foundry Models sold by Azure are billed via Azure meters as First Party Consumption Services" — states its own billing mechanism directly; independently confirmed.
- Name
- URL
-
Official documentation · 23 Sep 2026
Foundry Models' own product page: "Innovate faster using more than 11,000+ models packed for out-of-the-box-use"; serverless and managed-compute deployment. The old learn.microsoft.com url now lands on the '(classic)' portal docs.
Also from Microsoft
-
Microsoft Foundry Platform
Azure's platform for building, evaluating and deploying model-based applications and agents.
-
GitHub Copilot Developer tool
Coding assistant and agent inside the editor, the CLI and GitHub itself.
-
Microsoft 365 Copilot Application
AI assistant embedded in Word, Excel, PowerPoint, Outlook and Teams that grounds its answers in a customer's own business data and ships with pre-built agents.
-
Azure AI Search Data service
Enterprise retrieval service providing hybrid keyword and vector search plus RAG grounding over an organisation's own data, also surfaced inside Microsoft Foundry as Foundry IQ.
-
Microsoft Copilot Application
Microsoft's assistant for chat, search and work inside Windows and the Microsoft 365 apps.
-
Copilot Cowork AI agent
Agentic feature of Microsoft 365 Copilot that plans and executes multi-step, multi-app tasks end-to-end and returns a finished deliverable rather than a draft.
-
Microsoft Copilot Studio Agent platform
Low-code platform for building, connecting and publishing custom conversational and autonomous AI agents across Microsoft 365, the web and business channels.
-
Microsoft Security Copilot Application
Security-operations assistant that summarises threat signals and helps analysts detect, investigate and respond to incidents in natural language.
-
Azure AI Bot Service Application
Integrated development environment for building conversational bots, integrating with Microsoft Copilot Studio for low-code bot building.
-
Azure Content Understanding API service
Multimodal Azure AI service that extracts structured insights from unstructured content (documents, images, audio, video) into custom-defined schemas.
-
Azure AI Custom Vision API service
Image-recognition service for building custom image classification and object-detection models; distinct from the general-purpose Azure AI Vision service.
-
Azure Document Intelligence API service
Extracts text, structure and key-value pairs from documents (invoices, receipts, forms, ID cards) via prebuilt and custom models.
-
Azure AI Immersive Reader API service
Embeddable reading-comprehension tool that improves text readability for users of all ages and abilities.
-
Azure Language in Foundry Tools API service
NLP service for sentiment analysis, key phrase extraction, entity recognition, summarization, question answering and conversational language understanding.
-
Azure Machine Learning API service
End-to-end platform for building, training, deploying and managing machine-learning models at enterprise scale.
-
Azure OpenAI in Foundry Models API service
Hosted access to OpenAI's GPT, embedding and image models with Azure enterprise security, networking and compliance, under Microsoft Foundry's current branding.
-
Azure Speech in Foundry Tools API service
Speech-to-text, text-to-speech, speech translation and speaker recognition as an Azure API service.
-
Azure Translator in Foundry Tools API service
Real-time and batch text translation API supporting 100+ languages.
-
Azure AI Video Indexer API service
Extracts insights from video and audio — transcription, face/object detection, sentiment, topics — via API.
-
Azure Vision in Foundry Tools API service
Image analysis, OCR, spatial analysis and face detection as an Azure API service.
-
Microsoft Dragon Copilot API service
AI assistant for clinicians that combines ambient conversation capture, clinical documentation and voice-driven EHR/order workflows into one product.
-
Foundry Agent Service API service
Managed service for building, deploying and orchestrating enterprise AI agents with tool calling, state management and multi-turn conversations.
-
Content Safety in Foundry Control Plane API service
Detects harmful text/image content (hate, violence, sexual, self-harm) for generative-AI applications, billed per text/image record analyzed.
-
Foundry IQ API service
Context-engineering platform giving AI agents agentic retrieval across enterprise data sources (SharePoint, Blob Storage, the web) with permission-aware security.
-
Observability in Foundry Control Plane API service
Evaluation, production monitoring and distributed tracing for generative-AI applications, billed by consumption per the Foundry observability pricing page.
Open Microsoft in the directory
Something wrong here? Send a correction — quote this product id: microsoft-foundry-models.