Aider
Decision note
Aider is the open-source CLI pair-programmer maintained by Paul Gauthier, with native git integration that auto-commits each AI edit with a descriptive message. Aider supports bring-your-own-key against any major provider, runs on every platform (cli, on-prem, offline-friendly), and is the de facto reference for measuring agentic coding ability - most SWE-bench evaluations use Aider's evaluation setup. Free; you pay LLM API costs.
Overview
Open-source CLI tool for AI-driven code editing via natural language, with deep Git integration that automatically commits changes with descriptive messages.
Best LLMs for this agent
Aider's own leaderboard typically places Claude Sonnet 4.6 at the top for edit accuracy, and you can run it by setting the matching Anthropic API key.
Claude Opus 4.6 is the stronger Claude pairing for harder repo edits when accuracy is more important than token cost.
DeepSeek V4 Pro is a strong open-weights alternative on cost for agentic coding workflows where you still want frontier-class coding strength.
GPT-5 remains close behind the top Claude pairings for Aider and is available through the OpenAI API key path.
Pricing and Provider Comparison
Agent access tiers
| Tier | Price | Includes |
|---|---|---|
| Free / Open Source | Free | Open source, use any LLM via your own API keys |
Provider routes for recommended models
| Model | Provider | Input / 1M | Output / 1M | Notes |
|---|---|---|---|---|
| Claude Sonnet 4.6 | Anthropic | $3.00 | $15.00 | Batch pricing listed; Cache reads listed |
| Claude Opus 4.6 | Anthropic | $5.00 | $25.00 | Batch pricing listed; Cache reads listed |
| DeepSeek V4 Pro | DeepSeek Platform | $0.435 | $0.87 | Cache reads listed |
| GPT-5 | OpenAI API | $1.25 | $10.00 | Batch pricing listed; Cache reads listed |
Featured Model Specs
| Model | Lab | Released | Context | Reasoning | Openness |
|---|---|---|---|---|---|
| Claude Sonnet 4.6 | Anthropic | 2026-02-17 | 1m | Yes | Proprietary |
| Claude Opus 4.6 | Anthropic | 2026-02-05 | 1m | Yes | Proprietary |
| Mistral Large 3 675B Instruct | MistralAI | 2025-12-01 | 128k | No | Open source |
| Gemini 2.5 Pro | Google DeepMind | 2025-06-17 | 1m | Yes | Proprietary |
| GPT-4.1 | OpenAI | 2025-04-01 | 1.05m | No | Proprietary |
Model Compatibility Matrix
| Model | Lab | Default | Access tier | Notes |
|---|---|---|---|---|
| Claude Sonnet 4.6 | Anthropic | Yes | Free | Recommended default, via Anthropic API key |
| Claude Opus 4.6 | Anthropic | No | Free | Via Anthropic API key |
| Mistral Large 3 675B Instruct | MistralAI | No | Free | Via Mistral API key |
| Gemini 2.5 Pro | Google DeepMind | No | Free | Via Google API key |
| GPT-4.1 | OpenAI | No | Free | Via OpenAI API key |
| DeepSeek R1 | DeepSeek | No | Free | Via DeepSeek API key |
| DeepSeek V3 | DeepSeek | No | Free | Via DeepSeek API key |
| GPT-4o (2024-11-20) | OpenAI | No | Free | Via OpenAI API key |
Underlying Models
Capabilities
| Feature | Supported |
|---|---|
| Chat | Yes |
| Agent mode | Yes |
| Multi-file edit | Yes |
| Autocomplete | No |
| Terminal | No |
| Code review | No |
| Git integration | Yes |
| Deploy integration | No |
| Local models | Yes |
| Bring your own key | Yes |
| MCP support | No |
| Offline mode | Yes |
| IP indemnity | No |
| SSO | No |
| On-prem | Yes |