Azure OpenAI Service documentation

Azure OpenAI Service

Overview

• What is Azure OpenAI?
• Models
• Quotas & limits
• Programming languages/SDKs

What's new

• What's new in Azure OpenAI?

Quickstart

• Chat completions
• Vision-enabled models
• DALL-E
• Use your data (preview)
• Whisper
• Text to speech (preview)
• Assistants

Tutorial

• Embeddings
• Fine-tuning GPT-4o mini

Azure OpenAI concepts

Concept

• Deployment Types
• Quota
• Dynamic quota
• Provisioned Throughput Units (PTU)
• Content filtering
• Prompt engineering
• Fine-tuning
• Assistants

Azure OpenAI getting started

Get started

• Responses API & computer-use-preview
• Reasoning models
• OpenAI versus Azure OpenAI (Python)
• Global batch
• Structured outputs
• Chat completion models
• Vision-enabled models
• Provisioned Throughput Units (PTU)
• Network and access configuration for Azure OpenAI on your data
• Resource creation & deployment

Architecture & training

Architecture

• End-to-end chat reference architecture

Training

• Azure OpenAI training

Responsible AI

Reference

• Transparency note
• Limited access
• Code of conduct
• Data, privacy, and security
• Customer Copyright Commitment

Concept

• Asynchronous content filtering
• Red teaming large language models (LLMs)
• Safety system messages
• Abuse monitoring

Retrieval Augmented Generation (RAG) templates

Deploy

• C#
• Java
• JavaScript
• Python