Aider Review

Aider lets you pair program with LLMs to start a new project or build on an existing one. The README lists its main capabilities: working with cloud and local LLMs (it works best with strong models such as Claude Sonnet, DeepSeek, and OpenAI models, but can connect to almost any), mapping your entire codebase to help in larger projects, support for more than 100 languages, automatic Git commits with sensible messages so you can use familiar tools to diff and undo AI changes, use inside your IDE by adding comments to code, adding images and web pages to the chat, voice-to-code, automatic linting and testing after each change with the ability to fix detected problems, and a copy and paste mode for working with any LLM through its web chat. The README also shows badges claiming 6.8 million installs, 15 billion tokens per week, and a top-20 position on OpenRouter, and a Singularity statistic of 88%, which are self-reported project figures. The model names in the README's examples (Claude 3.7 Sonnet, o3-mini) date the documentation.

Install with python -m pip install aider-install and then run aider-install; change into your project directory and start aider with a model flag and an API key, for example aider --model sonnet --api-key anthropic=<key>, or a DeepSeek or OpenAI model. The installation and usage documentation lives on aider.chat. We did not run it.

Everything happens in the terminal: you add files to the chat, describe a change, and Aider proposes and applies edits, then commits them. Because each change becomes a Git commit, undoing a bad edit is a normal git operation. We did not test it hands-on.

• Terminal pair programming with LLMs
• Repository map for larger codebases
• Automatic Git commits for every change
• Linting and testing after edits
• Works with cloud and local models
• Voice-to-code and image or web page context
• IDE integration through code comments

• Git-centric workflow makes AI edits reviewable and reversible
• Provider-agnostic, so you can choose the model and price
• Mature, widely used, about 49,600 stars
• Simple installation via a Python package

• The last push was in May 2026, so development has slowed compared with newer agentic tools
• README examples reference older model names, so expect to check current model recommendations
• Terminal-only, without a graphical interface
• Quality and cost depend on the model you connect

Aider is free under Apache-2.0. You pay for the model API you use, or run local models.

GitHub Copilot is a subscription IDE assistant with completions, chat, and agent features integrated into editors. Aider is a terminal tool that works with any model you supply and commits every change to Git. You get model choice and transparency instead of tight editor integration.

• Developers who live in the terminal and Git
• Teams that want to choose and swap models
• Anyone who wants AI edits as reviewable commits

• Users who want inline editor completions
• People who want an actively fast-moving agent framework
• Those unwilling to manage API keys