The video explains how developers can enhance Visual Studio Code and GitHub Copilot by integrating external AI models using the “bring your own key” (BYOK) approach, demonstrated with OpenRouter, and by utilizing the Foundry toolkit extension for accessing a wide range of AI models, including local deployments. These methods provide greater flexibility, control, and customization for coding workflows beyond the default Copilot offerings.
In this video, the presenter introduces the concept of “bring your own key” (BYOK) for Visual Studio Code and GitHub Copilot, explaining how users can integrate external AI models into their development environment. Traditionally, GitHub Copilot relies on proprietary models, but BYOK allows developers to use models hosted elsewhere by importing their API keys. This flexibility is beneficial for reasons such as organizational compliance, cost management, and performance optimization.
The presenter demonstrates the process using OpenRouter as an example. After obtaining an API key from OpenRouter, they search for a free model called DeepSeek v4. They then show how to configure Visual Studio Code by opening the Copilot chat window, accessing the language model selector, and adding a custom endpoint with the OpenRouter API key. By editing the chat language models.json file, they input necessary details like model ID, name, URL, and token limits, enabling the integration of the third-party model into the Copilot interface.
Once configured, the new model appears in the model selection dropdown within Visual Studio Code, allowing users to interact with it directly through Copilot chat. This setup effectively routes requests to the external OpenRouter model, demonstrating a practical way to extend Copilot’s capabilities beyond its default offerings. The presenter emphasizes that this method is just one approach to incorporating third-party AI models.
Additionally, the video explores another method using the Foundry toolkit extension available in the Visual Studio Code marketplace. This extension connects to Microsoft Foundry instances on Azure and provides access to a broad catalog of AI models, including those that can be hosted locally. The presenter highlights the ease of discovering, filtering, and using various models through this extension, which supports local deployment and integration within the development environment.
In conclusion, the video showcases how developers can enhance their coding workflows by bringing their own AI keys and models into Visual Studio Code and GitHub Copilot. Whether through direct API key integration or using extensions like Foundry toolkit, these options offer greater control, flexibility, and access to diverse AI capabilities. The presenter encourages viewers to engage with the content and experiment with these tools to improve their development experience.