Tabby

Tabby is a self-hosted AI coding assistant positioned as an open-source, on-premises alternative to GitHub Copilot. It runs as a single self-contained server — no external database or cloud service needed — and works with consumer-grade GPUs.

It provides code completion and a chat assistant through VS Code and JetBrains extensions, exposes an OpenAPI interface for integration with other IDEs and infrastructure, and can index your repositories (including GitLab merge requests) for context. The core is Apache-2.0; TabbyML also sells an enterprise edition with extra features.

1. Run the server with Docker: docker run -it --gpus all -p 8080:8080 -v $HOME/.tabby:/data tabbyml/tabby serve --model StarCoder-1B --device cuda.
2. Open http://localhost:8080 to finish setup and pick your models.
3. Install the Tabby extension in VS Code or JetBrains.
4. Point the extension at your server URL and start coding.