Deploying OpenAI GPT Model using Azure AI Foundry

This guide provides a comprehensive framework for deploying the OpenAI GPT model using Azure AI Foundry, making it accessible for users of varying expertise.

Hafeez Baig
14 steps
2 minutes

Azure

Refer to this guide to learn how to create hub and project in Azure AI Foundry - <https://scribehow.com/viewer/Exploring_Azure_AI_Foundry__Creating_Hub_and_Project__VIf2tVybTYCeZ6AOuinV4g>

On the **Azure AI Foundry** page, click on the resource project which you created in the previous lab

**Overview** page will open, click on the **Playgrounds** from the left bar

**Playgrounds** page will open, click on the **Try the Chat playground** button

**Deploy a model** widget will open, click on the **gpt-4.1** or whichever is the latest model

Then click on the **Confirm** button on the bottom right

**Deployment details** wizard will open, click on the **Deploy** button

**Chat playground** page will open, here you can click on any sample prompt

The sample prompt will be pre-populated, click on the **send** button the the bottom right

You will get the response from the AI Model

On the left side you can customize model instructions and context

**What is Model Instructions and Context?** **Model Instructions and Context** refer to the guidance and background information provided to an AI model to shape its behavior and responses. - **Instructions** tell the model *how* to respond, for example, "respond politely" or "summarize the input." - **Context** provides relevant background, such as previous messages, user preferences, or domain-specific details, helping the model generate more accurate and relevant outputs. Together, they ensure the model behaves consistently and understands the task it’s solving.

Below you will find **Parameters** dropdown, click on the dropdown option

You can customize the **Parameters** for your model here

**What are Parameters for AI Model?** **Parameters for an AI Model** are configurable settings that influence how the model generates its outputs. Common parameters include: - **Temperature**: Controls randomness; higher values produce more creative responses. - **Top-p (nucleus sampling)**: Limits the output to a subset of likely words for more focused responses. - **Max tokens**: Defines the maximum length of the response. - **Frequency and presence penalties**: Adjust repetition and encourage diverse content. These parameters help tailor the model's behavior for different use cases.

Below the instructions section click on the Add section dropdown and click on **Safety system messages** option

**What is Safety system messages for AI Model?** Safety system messages for an AI model are predefined instructions that guide the model to behave responsibly and avoid generating harmful, biased, or inappropriate content. These messages set boundaries on topics, tone, and allowed behaviors, helping ensure outputs align with ethical standards, compliance requirements, and organizational policies.

**Safety system messages** wizard will open, here you can customize and modify the safety systems for your AI Model

Congratulations on completing this lab and playing with chat playground!