Can Aider use Claude models?

Yes. Aider supports any LLM including Anthropic Claude — you supply your own API key. This also means you pay Anthropic API rates directly rather than through a seat subscription.

Does Aider support MCP?

MCP is not Aider's focus — its core value is git integration and tree-sitter repo mapping. Whether Aider has added MCP support since this page was written is unconfirmed; check aider.chat for the current status.

What happens to Claude Code pricing after 2026-06-22?

Claude Code is moving from being included in Pro/Max/Team plans to a usage-credits model. The exact rates were not confirmed at time of writing — check Anthropic's pricing page for current details.

How does Aider's auto-commit feature work?

After each AI-driven edit, Aider automatically stages and commits the changes to your local git repository. This gives you a clean, reviewable history of every AI change and makes it easy to revert any individual edit with a standard git revert.

Does Claude Code have a repo-map like Aider?

Claude Code reads files and runs shell commands as part of its agentic loop, but it does not have a dedicated pre-built tree-sitter repo map. Aider's repo map is a deliberate architectural choice to give the model precise multi-file context without burning tokens on irrelevant code.

Comparison

Aider vs Claude Code

Aider auto-commits every edit, builds a tree-sitter repo map, and works with any LLM you supply — all from a plain terminal, fully open source. Claude Code is Anthropic's proprietary terminal agent with MCP connectors, subagents, and hooks, locked to Claude models but deeply integrated with the broader Anthropic ecosystem.

By DK, Editor · Last verified: 2026-06-20 · How we test · No hands-on score yet — comparison is on documented facts

At a glance

Dimension	Aider	Claude Code
Form factor	Terminal CLI, open source	Terminal-native agent; also IDE plug-in and desktop app
Open source	Yes — fully open source (confirm license on aider.chat)	No — proprietary (Anthropic)
Free tier	Free to install; BYO model API key (no seat cost)	Included on Claude Pro/Max/Team plans until 2026-06-22, then moves to usage credits (confirm current)
Models / providers	Any LLM — Claude, GPT, local models, etc. (BYO API key)	Anthropic Claude only (Opus 4.8, Sonnet, Haiku)
MCP support	Not its focus — git + repo-map is the core; confirm current MCP status on aider.chat	MCP-native via connectors, subagents, hooks, skills, and plugin layers
Git integration	First-class: auto-commits each edit; architect/editor mode for large refactors	Can run git commands as part of multi-step tasks; no dedicated auto-commit layer
Repo map / context	Tree-sitter repo map across 100+ languages for precise multi-file targeting	Reads files and runs shell commands; context is agent-driven, not pre-mapped
Subagents / agent orchestration	No subagent/agent-team layer	Subagents and agent teams (native)
Best for	Developers wanting lightweight, model-agnostic, git-native multi-file refactors from the CLI	Developers in the Anthropic ecosystem needing a powerful agentic terminal with MCP and subagents

Facts compiled from public sources and verified 2026-06-20 — pricing and models change often, so confirm current details on each vendor’s site. No hands-on score is shown; this is a documented-fact comparison.

Verdict

If you want to own your model choice and keep a clean, automatic git trail on every AI edit, Aider is purpose-built for that workflow: install it with pip, point it at any LLM API key, and it will map your repo and commit changes for you. Claude Code fits a different mold — it is a full agentic runtime that can orchestrate subagents, connect to external services via MCP, and execute multi-step engineering tasks; the trade-off is single-vendor model lock-in and a pricing model that is shifting to usage credits after 2026-06-22 (confirm current terms). Neither tool is universally better: the decision turns on whether model portability and open-source auditability matter more to you than MCP integrations and Anthropic's agent-layer features.

Choose Aider if you want to BYO model (local, OpenAI, Claude, or anything else), keep costs fully transparent through your own API keys, and want every AI change automatically committed to git.
Choose Claude Code if you are already in the Anthropic ecosystem (Claude Pro/Max/Team), need MCP connectors to hook into external services, or want native subagent and agent-team orchestration for complex multi-step engineering tasks.
Choose Aider if open-source auditability is a hard requirement — you can read, fork, and audit every line of Aider; Claude Code is proprietary.
Choose Claude Code if IDE integration or desktop app access matters alongside your terminal workflow; Aider is a pure CLI with no editor GUI layer.

Our AI agents

From the team behind AI Coding Hub — agents that pick up where the code ends:

FAQ

Can Aider use Claude models?: Yes. Aider supports any LLM including Anthropic Claude — you supply your own API key. This also means you pay Anthropic API rates directly rather than through a seat subscription.
Does Aider support MCP?: MCP is not Aider's focus — its core value is git integration and tree-sitter repo mapping. Whether Aider has added MCP support since this page was written is unconfirmed; check aider.chat for the current status.
What happens to Claude Code pricing after 2026-06-22?: Claude Code is moving from being included in Pro/Max/Team plans to a usage-credits model. The exact rates were not confirmed at time of writing — check Anthropic's pricing page for current details.
How does Aider's auto-commit feature work?: After each AI-driven edit, Aider automatically stages and commits the changes to your local git repository. This gives you a clean, reviewable history of every AI change and makes it easy to revert any individual edit with a standard git revert.
Does Claude Code have a repo-map like Aider?: Claude Code reads files and runs shell commands as part of its agentic loop, but it does not have a dedicated pre-built tree-sitter repo map. Aider's repo map is a deliberate architectural choice to give the model precise multi-file context without burning tokens on irrelevant code.