Get started with Aider, the open-source AI pair programming tool. This tutorial covers installation, API key setup, basic usage, multi-file editing, and how Aider compares with Claude Code and Cursor.
They solve different problems. Aider is model-agnostic, open source, and works with any LLM provider. Claude Code has deeper Claude integration with tool use, MCP servers, and the full Claude Code harness. For pure pair programming, Aider is lighter. For complex multi-step agent workflows, Claude Code is more capable.
Yes, Aider supports Ollama, LM Studio, and any OpenAI-compatible endpoint. Use <code>aider --model ollama/llama3.3</code> or <code>aider --model openai-compatible/deepseek --openai-api-base http://localhost:1234/v1</code>. Performance varies by model size — larger models (70B+) work better for complex coding tasks.
Aider itself is free and open source (Apache 2.0). You only pay for LLM API usage. With Claude Sonnet, expect $5-15/month for casual use or $50-200/month for heavy daily coding. You can reduce costs by using cheaper models for simple tasks.
Aider uses a repo map to understand your project structure, but it only reads files you explicitly <code>/add</code> to the chat. Your code stays on your machine — Aider sends prompts and file contents to your chosen LLM provider's API. No code is sent to Aider's servers.
Aider works with any programming language since it reads and writes files as text. It has been benchmark-tested on Python, JavaScript, TypeScript, Go, Rust, Java, C++, PHP, Ruby, and more. Multi-language projects are supported natively.