Opilot VS Code Extension Integrates Ollama Models

What changed

The Opilot extension for VS Code has been updated to seamlessly integrate the Ollama ecosystem into GitHub Copilot Chat. This allows users to run any model from the Ollama library, including cloud-hosted models after authentication, as first-class Copilot chat models and as a dedicated @ollama participant. A new Model Management Sidebar provides tools to pull, run, inspect, stop, and delete models, complete with live status indicators. Users can now configure specific parameters for each model, such as temperature, top-p/top-k, context window size, and thinking budget, through an in-editor webview that persists settings per model. A status bar heartbeat visually confirms Ollama server connectivity. The extension also introduces inline code completions powered by local Ollama models and full support for tool calling and vision capabilities for compatible models. Furthermore, a Modelfile Manager facilitates the creation and editing of custom Ollama modelfiles with syntax highlighting, hover documentation, and autocomplete.

Why it matters for builders

This integration empowers developers by bringing the flexibility of local and cloud-based Ollama models directly into their coding workflow within VS Code. Builders can now experiment with diverse models without leaving their editor, enhancing privacy by keeping local model conversations on their machine. The granular control over model parameters and the ability to create custom modelfiles offer deeper customization for agentic workflows and specialized tasks. Inline code completion and tool/vision support further streamline development by providing context-aware suggestions and enabling more sophisticated AI-assisted coding.

Practical impact

Developers can install Opilot from the VS Code Marketplace and, after ensuring Ollama is running locally or remotely, can select Ollama models directly from the Copilot Chat model picker or invoke them using the @ollama participant. The Model Management Sidebar allows for easy management of local and cloud models, including pulling new models from the Ollama library. For inline code completions, developers can set opilot.completionModel in VS Code settings to a suitable local model. The Modelfile Manager offers a guided experience for creating custom models, with syntax support for modelfile syntax. Configuration options, accessible via VS Code settings, allow for customization of the Ollama server host, log streaming, and refresh intervals.

Caveats and source limits

The provided source details the features and configuration of the Opilot extension but does not include specific benchmark results comparing its performance against other integrations or native Copilot features. Pricing for any cloud-hosted Ollama models is not detailed, nor are specific performance metrics for local model execution. The source mentions a "fresh release" but does not provide a specific release date beyond a publication timestamp for the repository metadata. The exact set of compatible models for advanced features like tool calling and vision is not exhaustively listed, beyond mentioning "compatible models."