Today I want to show you how to use the Vision for Copilot Preview extension to add vision to your Github Copilot chat in VS Code. Let’s dive in…
We start by installing the Vision for Copilot Preview extension in VSCode:
Remark: This extension will be eventually deprecated in favor of built-in image flow in Github Copilot Chat.
At the moment of writing, the extension doesn’t use the existing model integration yet. Instead you need to configure a connection with either OpenAI, Azure OpenAI, Anthropic, or Gemini yourself.
Open the Azure Portal and go to https://ai.azure.com/. Click on Deployments:
Select one of your deployments and copy the full target URI:
Now go to the extension settings in VS Code for the Vision extension and paste the URI to your Azure OpenAI endpoint:
We also need to paste in the corresponding API key. Therefore, run the "Copilot Vision: Set Current Model's API Key" command and paste the key from Azure OpenAI Studio:
Once that is done, we can paste/drag & drop an image in our Copilot chat window and ask a question through @vision:
Here is the example output I got:
Nice!
More information
microsoft/vscode-copilot-vision: Exploration into leveraging vision capabilities of an LLM