How to Set Up Continue.dev with Ollama for Private AI Coding

Continue.dev is an open-source AI coding assistant for VS Code and JetBrains IDEs. It provides inline code completions, chat with your codebase, and the ability to edit files using natural language. Unlike GitHub Copilot, Continue.dev can connect to any model — including local ones via Ollama — ensuring your code never leaves your machine. It is free, privacy-focused, and highly configurable.

Before starting, ensure you have: Ollama installed (v0.5+ recommended) from ollama.com, VS Code (or any supported IDE), at least 8GB RAM (16GB+ recommended for larger models), and a code-optimized model downloaded. Continue.dev works with any Ollama model, but code-specific models yield the best results.

Download Ollama from ollama.com for your operating system. For Windows, run the installer executable. For macOS, download the .dmg or run "brew install ollama". For Linux, run the official install script: "curl -fsSL https://ollama.com/install.sh | sh". Verify installation by opening a terminal and running "ollama --version". You should see v0.5.0 or higher.

Continue.dev works best with models fine-tuned for code. Run "ollama pull codellama:7b-code" for a lightweight option that runs on most hardware. For better quality, use "ollama pull deepseek-coder:6.7b" or "ollama pull qwen2.5-coder:14b" for excellent multilingual coding support. For the best results on powerful hardware, try "ollama pull codellama:70b" or "ollama pull deepseek-coder:33b".

Open VS Code, go to the Extensions marketplace (Ctrl+Shift+X), and search for "Continue". Install the extension by Continue.dev. Alternatively, visit continue.dev to download the .vsix file for offline installation. Once installed, you will see a Continue panel on the left side of your editor.

Open Continue.dev settings by clicking the gear icon in the Continue panel. Under "Models", select "Ollama" as the provider. The default endpoint is http://localhost:11434. Choose your pulled model from the dropdown (e.g., codellama:7b-code). Set temperature to 0.2 for more deterministic completions. Enable "Allow anonymous telemetry" only if you want to help improve the tool — your code is never sent.

Continue.dev offers three main features. Autocomplete: start typing and suggestions appear inline — press Tab to accept. Chat: open the Continue panel and ask questions about your code, errors, or architecture. Edit: highlight code and describe changes in natural language (e.g., "refactor this into a React hook"). All processing happens locally through Ollama.

If suggestions are slow: switch to a smaller model like codellama:7b-code or enable GPU acceleration. If no suggestions appear: verify Ollama is running (check for the Ollama icon in your system tray or run "ollama serve"). If responses are irrelevant: adjust the system prompt in Continue.dev settings to specify your tech stack (e.g., "You are a TypeScript and React expert"). If you get out-of-memory errors: close other applications or use a model with fewer parameters. For setting up a more advanced agent system, see our guide on Hermes Agent with Ollama.