How to run your own self-hosted AI coding assistant

You can have code suggestions in your editor, like Copilot, but using a model that runs on your server. Your code never leaves your infrastructure.

What you need

• A local AI model specialized in code (for example through Ollama).
• The Continue extension in your editor (VS Code or JetBrains).

1. Run a code model

ollama run qwen2.5-coder

2. Configure the Continue extension

In the extension settings, point it to the local Ollama model as the provider. The extension will use your server for completions and code chat.

3. Run on a server or locally

You can run the model on a GPU server and connect from your laptop, so suggestions are fast without loading your laptop. See the GPU plans.

Advantages

• Confidential code, it does not go to an external service.
• Fixed cost, no monthly per-user subscription.
• You can pick the right model, see how to choose a model.

Combine with code-server for a complete development environment on the server.