El Bruno

> ⚠️ This blog post was created with the help of AI tools. Yes, I used a bit of magic from language models to organize my notes and automate the boring parts, but the geeky fun is 100% mine.

Hi!

The GitHub Copilot app can use models from your GitHub Copilot subscription, and it can also connect to models deployed in Microsoft Foundry.

This is a classic bring your own key, or BYOK, scenario:

GitHub Copilot app
        ↓
Custom Azure OpenAI provider
        ↓
Azure OpenAI v1 Responses API
        ↓
Model deployed in Microsoft Foundry

In this post, I will configure a gpt-5.6-sol deployment. The same pattern applies to other Azure OpenAI models that support the v1 Responses API.

Let’s do it! 🚀

TL;DR

This is the provider configuration:

This is the model configuration:

The two identifiers that matter are:

• Endpoint: identifies the Azure OpenAI resource and API surface.
• Wire model: identifies the deployment inside that resource.

Prerequisites

Before opening the GitHub Copilot app settings, you need:

1. A Microsoft Foundry or Azure OpenAI resource.
2. A deployed model that supports the Azure OpenAI v1 API.
3. The deployment name.
4. The resource endpoint.
5. An API key for the resource.
6. Network access from your computer to the Azure endpoint.

For this example:

Model:            gpt-5.6-sol
Deployment name:  gpt-5.6-sol
Deployment type:  GlobalStandard
Wire API:         Responses

Step 1: open the model deployment in Microsoft Foundry

Open Microsoft Foundry, select your project or resource, and navigate to:

Models + endpoints →  → Details

The deployment details page contains:

• the model and deployment name;
• the deployment type;
• the inference endpoint;
• authentication information.

The endpoint displayed by Foundry can be the complete operation URL:

https://.services.ai.azure.com/openai/v1/responses

For the GitHub Copilot app provider, use the base v1 endpoint:

https://.services.ai.azure.com/openai/v1/

The app adds the Responses operation based on the selected Wire API.

Resource endpoint versus operation URL

These values represent different layers:

Resource host
https://.services.ai.azure.com

API base path
/openai/v1/

Operation
responses

Combined, they produce:

https://.services.ai.azure.com/openai/v1/responses

The provider stores the first two parts. The selected Wire API provides the last part.

Step 2: add an Azure OpenAI provider

In the GitHub Copilot app, open:

Settings → Model providers → Add provider → Azure OpenAI

Use:

Display name:   Azure OpenAI
Endpoint:       https://.services.ai.azure.com/openai/v1/
API version:    v1
Wire API:       responses
API key:        
Custom headers: {}

Endpoint

The endpoint must:

• use the exact hostname assigned to the resource;
• include /openai/v1/;
• stop at the v1 base path;
• not include the deployment name;
• not include the API key;
• not include /responses.

Both Azure OpenAI endpoint formats are supported by the v1 API:

https://.openai.azure.com/openai/v1/

and:

https://.services.ai.azure.com/openai/v1/

Use the hostname shown for your resource in Microsoft Foundry or the Azure portal.

API version

Use:

v1

The Azure OpenAI v1 API is a continuously updated API surface. Unlike older Azure OpenAI APIs, it does not require applications to move between monthly dated API versions.

In the GitHub Copilot app provider UI, v1 selects the correct API generation for this configuration.

Do not confuse the API version with the model version:

Wire API

Select:

responses

This tells the app to use:

POST /openai/v1/responses

and the Responses API request schema.

Conceptually, the request body is:

{
  "model": "gpt-5.6-sol",
  "input": "Hi!"
}

This is different from the older Chat Completions API, which uses a messages array.

API key

Paste one of the keys associated with the same Azure resource as the endpoint.

The important relationship is:

API key resource == endpoint resource

Do not add the key to custom headers or include it in the endpoint.

Custom headers

No custom headers are required for standard inference with the v1 Responses API:

{}

Some feature-specific preview APIs use opt-in headers. Basic text generation with a deployed model does not need them.

Step 3: add the deployed model

Inside the new Azure OpenAI provider, select Add model.

Configure:

Display name:      gpt-5.6-sol
Wire model:        
Reasoning effort:  High

The provider status indicator should be healthy.

At this point, the app has enough information to construct the complete request:

Provider endpoint
https://.services.ai.azure.com/openai/v1/

Wire API
responses

Wire model
gpt-5.6-sol

Authentication
Azure OpenAI resource key

Step 5: select the model and send a prompt

Create a new chat and select the newly configured Azure OpenAI model.

Send a simple prompt:

Hi!

Final configuration

The final setup is:

Microsoft Foundry deployment
        ↓ deployment name

GitHub Copilot app model
        ↓ Wire model

Azure OpenAI provider
        ├── Endpoint: .../openai/v1/
        ├── API version: v1
        ├── Wire API: responses
        └── Authentication: resource API key

This gives the GitHub Copilot app direct access to your Azure-hosted deployment while keeping model routing, quota, and billing attached to your Microsoft Foundry resource.

Simple, explicit, and ready for more models. 😁

References

• Azure OpenAI v1 API lifecycle
• Use the Azure OpenAI Responses API
• Microsoft Foundry