Free & Open Source
$0
MIT/Apache 2.0, BYOK, pay only LLM API costs
Aider
Open-source AI pair programmer in your terminal — Git-native, model-agnostic.
Aider is the leading open-source AI pair programmer for the terminal (~39K GitHub stars, 4M+ installs). It maps your entire codebase, lets you pair-program with any LLM (Claude, GPT-5, Gemini, DeepSeek, Grok, Mistral, or any local model) through a natural-language chat, and auto-commits every change to Git with a descriptive message — making AI edits fully auditable and reversible. Architect/editor mode pairs a strong reasoner with a fast executor for cheaper, more reliable multi-file work.
+ Pros
- Supports 100+ programming languages and connects to virtually any LLM — from cloud APIs (Claude, GPT-4o, DeepSeek) to local models via Ollama, with automatic codebase mapping for context in large projects.
- Built-in Git integration that automatically commits AI-generated changes with sensible commit messages, making it easy to diff, review, and undo modifications using familiar Git workflows.
- Auto-linting and testing on every change — Aider runs your linters and test suite after each edit and can self-correct problems detected, reducing manual review cycles.
- Voice-to-code capability lets you speak feature requests, test cases, or bug fixes directly, with Aider implementing the changes from your spoken input.
- Open source (Apache-2.0), free to use, with 45.4k GitHub stars, 6.8M+ PyPI installs, and an active community — no subscription or vendor lock-in required.
- Architect/Editor dual-model mode: pairs a strong reasoning model to design solutions with a cheaper, faster model to implement them, optimizing both quality and cost.
− Cons
- Terminal-only experience — no native IDE plugin or graphical UI; users must be comfortable with CLI tools and command-line workflows.
- Requires external API keys for best-in-class models (Claude, GPT-4o, etc.) with per-token costs; quality degrades significantly with free or weaker local models.
- Steep learning curve for advanced configuration — understanding edit formats, model aliases, architect mode, and YAML configuration can overwhelm new users.
- Repository map generation and context handling can be slow on very large codebases, especially during initial startup and when working with monorepos.
Pricing
Aider is the leading open-source AI pair programmer for the terminal, created by Paul Gauthier and distributed under the Apache 2.0 license. Unlike IDE-based assistants, Aider runs in your CLI and works alongside any editor — VS Code, Vim, Neovim, Emacs, or no editor at all. Its core innovation is the repo map: Aider automatically builds an internal map of the entire codebase so every prompt is grounded in real project context, even for large monorepos across 100+ programming languages. Paul Gauthier also maintains the Aider Polyglot benchmark on 225 Exercism problems, widely cited as a reference signal for raw model coding capability.
Aider is completely model-agnostic and BYOK: it connects to any cloud LLM (Claude Opus 4.7 / Sonnet 4.6, GPT-5 family, Gemini 2.5 Pro, DeepSeek V3.2 / R1, xAI Grok 4, Mistral) or any local model via Ollama and LM Studio. You pay provider API rates directly with zero markup. A key privacy property: only the code relevant to the current prompt is sent to the LLM, so the rest of your source never leaves your machine. Every AI-generated change is automatically staged and committed to Git with a descriptive commit message, so all AI edits are fully auditable and reversible using standard Git tools — rollback is just `git reset`.
Aider's signature 2026 features are its architect/editor mode (a stronger reasoning model plans the changes, a faster/cheaper model applies them — measurably better on multi-file work, typically 30–50% cheaper than running the architect alone), watch mode with `AI!` comments (drop `# AI! refactor this to use async`-style comments anywhere in your code and Aider picks them up automatically), explicit `/code`, `/ask`, and `/architect` mode switching mid-session, voice-to-code, image and web page context, automated linting and test execution with auto-fix, and a `.aider.conf.yml` that turns AI-coding policy into version-controlled team practice.
Key Features
• Repo map: full codebase context across 100+ languages
• Model-agnostic: Claude, GPT-5, Gemini, DeepSeek, Grok, Mistral, plus local via Ollama / LM Studio
• BYOK at provider rates — no markup, no subscription
• Git auto-commit: every change is an atomic commit with a descriptive message
• Architect/editor mode: planner model + executor model, 30–50% cheaper than architect alone
• Modes: /code (default), /ask (read-only Q&A), /architect (plan first), /web (URL scrape), /run (shell), /read-only
• Watch mode with `AI!` comments: drop natural-language comments and Aider picks them up
• Voice-to-code: speak prompts to add features, fix bugs, write tests
• Image & web context: include screenshots or reference URLs
• Lint & test auto-fix: runs your tools after each edit and iterates on failures
• Prompt caching, reasoning-effort and thinking-token flags for frontier models
• .aider.conf.yml: version-controlled, team-shareable agent configuration
• Privacy: only relevant code is sent to the LLM
• Aider Polyglot benchmark (Paul Gauthier) cited across the industry
• ~39K GitHub stars · 4M+ installs · Apache 2.0
Version History
Added support for all GPT-5 model variants, Grok-4 via xai/grok-4, Gemini 2.5 Flash Lite preview, and Kimi K2 model. Aider wrote 88% of this release's code.
Support for new Gemini 2.5 Pro/Flash models with thinking tokens, o1-pro and o3-pro via Responses API, co-authored-by commit attribution by default, MATLAB and Clojure repo-map support.
Added Claude Sonnet 4 and Opus 4 series support across providers, auto-refresh for GitHub Copilot tokens, improved OpenRouter model metadata caching and tab completion for file paths.
Support for Qwen3-235b and Gemini 2.5 Pro preview, OCaml repo-map, new spinner animation, dropped Python 3.9, auto-fetch OpenRouter model params, and --shell-completions flag.
GPT-4.1, mini and nano support with patch edit format, Grok-3 beta models, Optimus Alpha alias, new editor-diff/editor-whole edit formats, and architect mode improvements for Gemini 2.5 Pro.
Signature Snippet
In the terminal: `aider --model claude-sonnet-4-5 src/auth.py`. Aider maps the repo, then you type: 'Add rate limiting to the login endpoint, write a test, and update the docs.' Aider plans the changes across 3 files, applies the edits, runs the linter, commits each step to Git with descriptive messages — all without leaving the terminal.More in this Space
Vix
Vix is a Go-native, open-source (AGPL-3.0) AI coding agent that slashes token costs by 40-50% using a stem agent architecture and Tree-sitter virtual filesystem. It rethinks the plan/execute loop — keeping LLM cache warm across Explore, Plan, and Execute phases — while shipping Programmable Workflows, Whiteboard Mode with voice AI, MCP server support, and a self-evolving agent that writes its own scheduled jobs and watchers.
Late — High-Leverage AI Agent Orchestration
Orchestrate an entire AI dev team on 5GB VRAM using ephemeral subagents, exact-match diffs, and a zero-dependency Go binary. Works with any OpenAI-compatible model — local or cloud.
Paca
Paca is a free, open-source, self-hosted Scrum board where AI agents work as equal teammates — assigned to sprints, picking up tasks, and collaborating on BDD specs alongside humans. Built as an alternative to Jira and Linear, it treats AI agents as first-class Scrum members.