Aider vs Continue.dev — AI Coding Assistants Compared

AI-assisted coding has moved beyond experiments into daily workflow for millions of developers. But not all AI coding tools work the same way — and the differences matter more than which model they use underneath.

A recent discussion on Lobsters framed the divide as “plotting vs pantsing” — planning your solution ahead of time versus iterating your way to one. The same divide exists in AI coding tools: some act as delegates (you describe the goal, they build it autonomously) while others act as assistants (they collaborate, suggest, and complete as you work).

Two open source tools represent these approaches better than any: Aider and Continue.dev. Both are free, open source, and widely used. Both connect to local or cloud LLMs. But they serve very different workflows. Let’s break down how they compare.

Aider — The AI Delegate

Aider is an AI pair programming tool that runs entirely in your terminal. With 47,500+ stars on GitHub and an Apache 2.0 licence, it’s one of the most popular open source AI coding tools available.

The core idea is simple: you describe what you want in natural language, and Aider writes the code. It reads your codebase, makes a map of your project structure, and edits files directly. It then automatically commits changes with sensible commit messages — so every AI-generated change is tracked in git.

Key features:

• Codebase-aware editing — Aider maps your entire project structure, so it understands how files relate to each other before making changes.
• 100+ language support — Python, JavaScript, Rust, Go, C++, and dozens more. It reads your project’s file extensions to determine context.
• Automatic git integration — Every change gets a commit with a descriptive message. You can diff, revert, or squash using familiar git commands.
• Voice-to-code — You can speak your requests and Aider implements them.
• Linting and testing — Aider automatically runs your linters and test suites after each change, fixing problems it detects.
• Works with cloud and local LLMs — Claude 3.7 Sonnet, DeepSeek, GPT-4o, o3-mini, and dozens of local models via Ollama or LM Studio.

Aider’s “delegate” nature makes it especially powerful for experienced developers who know what they want and just need the code written. You describe the specification, Aider implements it, and git tracks everything in case you need to undo.

Best for: Developers who work in the terminal, know exactly what they want built, and want AI to handle implementation while maintaining full git control.

Continue.dev — The AI Assistant

Continue takes a fundamentally different approach. It integrates directly into your IDE — VS Code, VS Codium, or JetBrains — and works with you rather than for you. With 35,000+ stars and an Apache 2.0 licence, it’s the leading open source AI coding assistant.

Think of Continue as an AI pair programmer who sits next to you, reads your code as you type, and offers suggestions, completions, and refactoring ideas in real time. It doesn’t take over files — it suggests edits that you review and accept.

Key features:

• IDE-native integration — Appears as a sidebar and inline panel in VS Code, VS Codium, and JetBrains IDEs. No context switching.
• Tab autocompletion — AI suggests the next few lines of code as you type, similar to GitHub Copilot but self-hosted.
• Inline editing — Highlight a block of code and ask Continue to refactor, document, optimise, or explain it. Changes appear as a diff you can accept or reject.
• Multi-model support — Works with Sonnet, DeepSeek, GPT-4o, local models, and any OpenAI-compatible API endpoint.
• Custom slash commands — Define your own prompts (e.g., /test to generate unit tests, /doc to add documentation).
• Context-aware chat — The assistant references your open files, selected code, and terminal output to provide relevant suggestions.

Continue’s 2.0.0 release marked a significant milestone — removing telemetry, pulling out authentication, and polishing the experience for fully offline, privacy-respecting use. The repository is now in read-only maintenance mode, but the 2.0 release is stable and complete.

Best for: Developers who want real-time AI assistance within their existing IDE workflow, prefer reviewing suggestions before applying them, and value the assistant-style collaborative approach.

Head-to-Head Comparison

Which One Should You Choose?

The right tool depends entirely on how you work:

Choose Aider if: You live in the terminal, prefer describing what you want and letting AI implement it, and want every change tracked in git. Aider is ideal for experienced developers tackling greenfield projects or adding features to existing codebases where you already understand the architecture — the “plotting” approach.

Choose Continue.dev if: You use VS Code or JetBrains as your daily driver, prefer reviewing AI suggestions before they touch your code, and value real-time autocomplete and inline assistance — the “pantsing” approach. Continue shines when you’re exploring new codebases, learning unfamiliar APIs, or iterating quickly on UI work.

Use both if: Many developers do. Continue handles the moment-to-moment suggestions while you’re actively typing, and Aider handles the larger feature-level implementation tasks. They complement rather than compete.

The important thing is that both are fully open source — Apache 2.0 licensed, no telemetry (in their final/current releases), and compatible with local LLMs for fully offline use. No monthly subscription required, no data leaving your machine.

Whether you’re a plotter or a pantser, there’s a free and open source AI coding tool that fits your workflow. Both Aider and Continue.dev are already listed in our directory, alongside many other free and open source development tools.