Aider.
Terminal-native pair programmer — git-aware, model-agnostic.
What it is.
Open-source CLI pair programmer. Written by the open-source community. Runs in the terminal, edits files in place, commits diffs to git automatically. Model-agnostic — point it at any OpenAI-compatible endpoint, including local Llama servers.
Where it fits.
Engineers who live in the terminal and want full control over the model, the cost, and the iteration loop. The benchmark-leading edit-rate scores Aider posts each month make it the reference for how good model-X is at code in practice.
- Open-source, scriptable, no vendor lock-in
- Git-aware — every edit committed atomically
- Public model benchmarks updated monthly
- Terminal-first UX excludes non-engineer collaborators
- No managed service — buyer runs all infrastructure
Frequently asked.
Why use Aider over Cursor?
Full transparency and control. Every model call, every diff, every commit is visible. Most useful for engineers running cost-sensitive experiments or training internal benchmarks.
Does Aider work with local models?
Yes. Point it at any OpenAI-compatible endpoint — Ollama, vLLM, llama.cpp server. Quality depends on the model.
What are the Aider benchmarks?
The polyglot benchmark — code-edit rate across multiple languages. Updated monthly. Treated as a reference scoreboard by the wider community.
