Azure OpenAI in Foundry Models provides API access to current OpenAI models, including the GPT-5.x series, the GPT-4.1 series, the o-series reasoning models, embeddings, and image, audio, and realtime models. You use these models with the security, compliance, and regional availability of Microsoft Azure.
What is Azure OpenAI in Foundry Models?
Azure OpenAI in Foundry Models is part of Microsoft Foundry (formerly Azure AI Foundry) and delivers powerful OpenAI models as a fully managed cloud service. You access reasoning, chat, embedding, image, and audio models through a unified API, without operating your own infrastructure.
The service combines OpenAI’s models with the Azure ecosystem: unified billing, governance, identity integration, and network security. This lets you build production AI applications that meet enterprise requirements for security and data protection.
The model line-up is updated very frequently: several interim versions (including GPT-5.1 through GPT-5.5) have followed the original GPT-5 series, along with specialized codex variants for developer workflows. For current model names and capabilities, the Foundry model documentation is the most reliable source.
For companies with GDPR requirements, operation within the EU Data Boundary is possible. With Data Zone deployments, processing of prompts and completions stays exclusively in EU regions.
Core Features
- Continuously updated model line-up: GPT-5.x series including codex variants, GPT-4.1 series, o-series reasoning models (o3, o3-pro, o4-mini), embeddings, image generation, and audio and realtime models.
- Large context windows: depending on model and deployment type, some exceeding 1M tokens for large documents and long conversations.
- Flexible deployment types: Standard and Data Zone (pay-per-token), Provisioned (reserved PTU capacity), and Batch for asynchronous bulk processing.
- Enterprise security: VNET integration, Private Link, managed identity, Azure Policy controls, and built-in content safety filters.
- Data residency and compliance: EU Data Zone within the EU Data Boundary, with no use of customer data for model training.
- Extensibility: function calling, structured outputs, the Responses API, fine-tuning for selected models, and integration with Azure AI Search for retrieval augmented generation.
Typical Use Cases
- Conversational AI, chatbots, and internal copilots for knowledge work and support.
- Content generation, translation, and summarization across large document sets.
- Code generation and developer assistance within existing workflows.
- Semantic search and RAG scenarios using embeddings and Azure AI Search.
- Document analysis, classification, and structured data extraction from unstructured sources.
Benefits
- GDPR-compliant operation through the EU Data Zone and processing within the EU Data Boundary.
- Predictable performance via Provisioned Throughput Units (PTU) with guaranteed throughput.
- A financially backed availability SLA for standard deployments.
- Cost optimization by choosing the right deployment type, including lower-cost Batch processing.
- Seamless integration into Azure: identity, monitoring, networking, and centralized governance.
Integration with innFactory
As a Microsoft Solutions Partner, innFactory supports you with the architecture, rollout, and operation of Azure OpenAI in Foundry Models. We help you select the right models and deployment types, build GDPR-compliant architectures in the EU Data Zone, design RAG solutions with Azure AI Search, and optimize cost through PTU and Batch.
Contact us for a non-binding consultation on Azure OpenAI in Foundry Models and Microsoft Azure.
Available Tiers & Options
GPT-5.x series
- Continuously updated reasoning models for complex tasks (currently including GPT-5.5, GPT-5.4, GPT-5.1 Codex variants)
- Text and image processing
- Large context windows (model-dependent, some over 400K tokens)
- Variants: mini, nano, pro, codex depending on generation
- Higher cost with heavy reasoning
- Frequent model updates require regular version checks
GPT-4.1 series
- Very large context window (over 1M tokens depending on deployment type)
- Cost-effective (mini, nano)
- Function calling and structured outputs
- No reasoning mode
- Older than the current GPT-5.x generation
o-series (reasoning)
- o3, o3-pro, and o4-mini for deep analysis
- Step-by-step reasoning
- Strong at code and math
- Higher latency and cost
Typical Use Cases
Technical Specifications
Frequently Asked Questions
Which models are available in Azure OpenAI in Foundry Models?
Available models include the continuously updated GPT-5.x line-up (including mini, nano, pro, and codex variants), the older but still supported GPT-4.1 series, o-series reasoning models (including o3, o3-pro, o4-mini), embeddings, image generation, and audio and realtime models. The model catalog is updated very frequently; check the Foundry model documentation for the current list.
How is Azure OpenAI different from OpenAI's API?
Azure OpenAI offers the same underlying models with added enterprise features: availability SLA, Azure security, VNET integration, managed identity, Azure Policy controls, and data residency in EU regions via the EU Data Zone.
Is my data used to train models?
No. Your prompts and completions are not used to train, retrain, or improve OpenAI or Microsoft models. Your data stays your data.
Is Azure OpenAI in Foundry Models GDPR compliant and EU-resident?
Yes. With Data Zone deployments in the EU, prompts and completions are processed only within the EU Data Boundary, which covers regions in countries such as Germany, France, the Netherlands, and Sweden. Microsoft also provides data processing agreements and comprehensive compliance certifications.
How does the pricing model work?
There are two billing models: pay-per-token usage (Standard, Data Zone) or reserved capacity via Provisioned Throughput Units (PTU) for predictable performance. For asynchronous bulk processing, Batch deployments offer reduced cost compared to standard deployments.
What SLA applies to Azure OpenAI?
For standard deployments, Microsoft publishes a financially backed availability SLA for the inference endpoint. Provisioned deployments additionally provide guaranteed throughput and lower latency variance. Check the official SLA page for exact percentages, as they can differ by deployment type.
Note: All product information on this page has been compiled with care, but is provided without guarantee and may be outdated or incomplete. Cloud services evolve rapidly — features, pricing, SLAs, and availability change frequently. Authoritative and up-to-date information can only be found on the official product page of Azure (official documentation). This page does not represent an offer by Azure.
