Aider

Terminal-based AI pair programmer (open source)

Code & Dev Tools Open Source Open Source
Researched · Published · Reviewed
RECATOOLS Score
8 / 10
Capability
8
Value for money
9
Ease of use
6
ASEAN readiness
7
API quality
6
Founded
2023
HQ
Seattle, Washington, USA
Users
100k+ developers
Launched
Jun 2023
Developer
Paul Gauthier

Overview

Aider is a free, open-source CLI tool (Apache 2.0, 47k+ stars) that turns Claude, GPT, DeepSeek, or local models into a pair programmer working directly against your Git repo, auto-committing edits. It adds an architect/editor split mode, voice-to-code, and IDE watch mode.

Advertisement

Pricing

Pricing shown for reference only. These figures reflect RECATOOLS research as of 22 Jul 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.

Free
Free
Fully free; only cost is LLM API usage

Use cases

Terminal coding Local LLM coding Git-aware edits

What you can produce with Aider

  • Multi-file edits made directly in your Git repo, with its repo map supplying the context you would otherwise paste in by hand
  • Every change landing as a Git commit with a generated message, so review and rollback are ordinary Git rather than a proprietary undo
  • Model choice at the command line — Claude, GPT-5, DeepSeek R1, or anything local through Ollama or LM Studio
  • Fully local operation with nothing leaving your machine, when run against a local model
  • Better accuracy on large refactors through the architect/editor split: one model reasons about the change, a second writes the diff
  • All of it free under Apache 2.0 on your own API keys — but it is a CLI with a real learning curve, still on 0.x versioning, so pin a version in CI rather than pulling latest
Advertisement
RECATOOLS Verdict

Aider is a command-line AI pair programmer that edits files directly in your local repo, builds a repo map for context, and commits changes through Git with generated messages — genuinely useful for reviewing diffs rather than trusting an opaque agent. It's model-agnostic: point it at Claude, GPT, DeepSeek, or a local model via your own API keys, so you control both cost and capability, and the architect/editor split mode (one model plans, another writes) is a real quality bump on hard tasks. It has a large, active community. The tradeoffs: it's terminal-first with a real learning curve, you pay token costs yourself, and it's still pre-1.0 software — breaking changes to CLI flags and the config format do happen between minor releases.

Independent AI-assisted assessment by RECATOOLS.

What people say

Aider remains one of the most widely adopted terminal-based coding agents, with reach well beyond most of its well-funded IDE competitors.

The core workflow hasn't changed much since launch: point Aider at a Git repo, pick a model (Claude, GPT-5, DeepSeek R1, or anything local via Ollama), and it edits files directly, running its own repo map so multi-file context doesn't need to be pasted in manually. Every change lands as a Git commit with a generated message, which makes review and rollback trivial compared with agents that mutate files silently. The newer architect/editor split — one model reasons about the change, a second one writes the diff — shows up often in community benchmarks as a meaningful accuracy gain over single-model editing, particularly on larger refactors.

It remains genuinely free and open source under Apache 2.0, with no hosted product to buy; you supply your own API keys and pay the underlying provider directly, which local-LLM users like because nothing leaves their machine if they run against Ollama or LM Studio. The catch is the same one it's always had: this is a CLI tool with a real learning curve, not a point-and-click assistant, and large architectural refactors still need a careful human pass before merging. The project is also still on 0.x versioning, so CLI flags and the .aider.conf.yml format have changed across minor releases — worth pinning a version in CI rather than always pulling latest.

Summary of public user & expert reviews, compiled by RECATOOLS.

Notable facts

  • Aider consistently ranks at the top of the SWE-Bench benchmark for multi-file code editing tasks, outperforming many commercial coding agents.
  • The tool automatically writes Git commit messages for every change it makes, including a description of what was changed and why.
  • Aider can work with an entire codebase context that exceeds 100,000 tokens by selectively including only the most relevant files for each task.

Frequently asked questions

Yes. Aider itself is free and open source. You pay for the LLM API (Claude, GPT-4) that powers it.
Claude, GPT-4, Gemini, and local models via Ollama. Claude models are generally recommended for best results.
Yes. Aider has no language restrictions and works on any codebase.
Aider maps the codebase structure and selectively adds only relevant files to context for each request.
Yes. Every change is automatically committed with a descriptive message. Reverting changes is a standard git revert.

About this listing

Researched on
Published on
Last reviewed

This entry was compiled from publicly available data including Aider's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Aider unless explicitly stated.

Data accuracy

Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.

For the latest details, please refer to Aider directly →

Spotted something out of date? Suggest an update →

Advertisement