Aider Pricing: Free Software and the Real Cost of APIs

Aider, the open-source coding assistant that runs in the terminal, costs €0. It is Apache-2.0 software, and aider.chat lists no plan, tier, seat, or hosted service: “Aider pricing” in practice means the model API bill for calls made through it. The figure to budget for is tokens per session multiplied by your provider’s rate—and the setting most likely to inflate that figure is Aider’s default behavior for a model ID it does not recognize. Two lines fix it; details below.

A clarification first, because search results mix the two: another company that publishes accounting close software for firms is also called Aider. The monthly prices it publishes are for that product. This page covers only the CLI from aider.chat, distributed as the aider-chat package from the Aider-AI/aider repository.

What Aider charges and what actually charges you

The Aider FAQ says the tool is published as open source under Apache 2.0; there is no mention of pricing, a plan, or a subscription on the homepage, FAQ, installation page, or LLM connection pages (verified October 1, 2026). The entire bill is on the other side of the endpoint.

Check the release cadence before standardizing on the tool across a team. The repository is not archived and has 49,309 stars, but the last commit on main was on May 22, 2026 (GitHub API, October 1, 2026). The two official sources for the “latest version” also disagree: PyPI serves 0.86.2, published on February 12, 2026, while GitHub Releases stops at v0.86.0, from August 9, 2025. It is PyPI that pip install installs. For Python, the constraint actually enforced by pip is the one in the 0.86.2 package metadata, >=3.10,<3.13, not the broader range on the installation page.

The setting that affects the API bill the most

This is the part pricing tables don’t show. Aider chooses its edit format, repo map, and cache behavior by comparing the model ID against a built-in table—and the ID being compared is the full string you type after --model, including the prefix. Its defaults when nothing matches are edit_format: whole, use_repo_map: false, cache_control: false, and weak_model_name: null, according to the advanced settings page.

In the 0.86.2 source code, lookup happens in two stages: first an exact match against 313 fully qualified names (for example, gemini/gemini-3-pro-preview or openai/gpt-5.2); only if that fails does it move on to substring rules that catch, among others, sonnet-4-5, haiku-4-5, or gpt-4.1. To connect an OpenAI-compatible gateway, the model ID must be given as openai/<id>, but all 40 entries prefixed with openai/ in the table are model names from OpenAI itself. So openai/claude-sonnet-4-6 matches nothing, and Aider uses whole-file rewriting, without a repo map: on every turn, the model returns each edited file in full. Output tokens are the expensive side of the bill.

Two lines fix this. Aider reads this file from your home directory, the git root, or the current directory. Its key is the full name passed to --model:

# name doit être identique à la chaîne passée à --model, préfixe compris
- name: openai/claude-sonnet-4-6
  edit_format: diff
  use_repo_map: true

For a one-off run, --edit-format diff is enough. By contrast, openai/claude-haiku-4-5 matches haiku-4-5 by substring and automatically gets diff format and the repo map—a naming coincidence that may change with any Aider release, so check it rather than assume it.

A second, subtler effect: Aider calls the API in three roles—the main model, an editor model used in --architect mode, and a weak model for commit messages and history summaries (see the options reference). With an unrecognized ID, the code uses the main model as both the weak and editor models, so all three roles are billed at the highest rate. --weak-model lets you specify a cheaper ID (with the same openai/ prefix).

Finally, an unrecognized ID hides the cost display too: without input_cost_per_token, Aider shows tokens but no cost line, and its warnings page confirms that it “assumes the model is free.” A metadata file restores it:

{
  "openai/claude-sonnet-4-6": {
    "input_cost_per_token": 0.0000021000,
    "output_cost_per_token": 0.0000105000,
    "litellm_provider": "openai",
    "mode": "chat"
  }
}

Even with this metadata, the amount Aider shows remains its own calculation based on reported token counts, with cache multipliers hard-coded for one specific provider. Treat it as an estimate and compare it with your provider’s usage record.

A worked estimate per session

The following is token arithmetic, not a measured task cost or invoice ceiling. Assume a session sends 200,000 uncached input tokens and receives 12,000 output tokens in diff format. For the whole-file column, assume the same work produces 40,000 output tokens because edited files are returned in full—an illustrative ratio, not a measurement. Rates are the current prices in the Kunavo catalog per million tokens.

With these assumptions, the edit format alone changes Claude Sonnet 4.6 for the same work from $0.546 to $0.840. This is why you should fix the setting before comparing provider rates. Commit message and summary traffic (assume 30,000 input and 2,000 output tokens) costs $0.084 on Claude Sonnet 4.6 if it stays with the main model, versus $0.028 if you specify Claude Haiku 4.5 as the weak model.

Multiply by your number of daily sessions before treating this as a budget. Kunavo’s catalog price is a billing floor, not a ceiling. When the upstream provider reports a cost, the charge is the greater of the catalog price and the upstream cost multiplied by the applicable markup. Cache costs and external tools are not included in this example.

“The cheapest API” is two different questions

Aider publishes the only task-cost benchmark for this tool that can be checked again: its polyglot benchmark has each model solve the same 225 Exercism exercises in six languages and records the cost in dollars.

Values read from the leaderboard data file. Two caveats matter more than the ranking: the latest row is dated October 3, 2025, so nothing released since then has been measured; and the cost column gives neither a pricing schedule nor a date—it is what that run cost that day, not a price available for purchase today. These models are cited as benchmark evidence; DeepSeek is not in the Kunavo catalog, and the inexpensive counterpart in the previous table is Claude Haiku 4.5.

The useful takeaway: o3-pro cost five times as much as gpt-5 (high) for a lower score, so a high rate does not buy completion; and a low rate does not guarantee the lowest total either, because a cheap model that needs three retries can cost more than an expensive model that succeeds on the first try. Pick the cheapest model that completes your work with an acceptable review effort, then check the amount actually recorded in your account for that task. Aider’s /tokens measures, as defined in its reference, the context of the current conversation, not the total already billed.

Which route to choose, and when

Two limits on the gateway route. A gateway without a dedicated module connects through the compatibility route documented on the OpenAI-compatible APIs page: OPENAI_API_BASE, OPENAI_API_KEY, and the required openai/ prefix. No Anthropic-format base URL override is documented, and the string ANTHROPIC_API_BASE does not appear in the 0.86.2 source code. Also, do not count on a cache discount on this route: the cache page limits support to certain provider families, requires the --cache-prompts option, and says cache statistics and costs are not shown while streaming. This page has not tested whether cache works through the compatibility route.

Using Aider with Kunavo and paying from France

Kunavo publishes a setup guide for this client. It is a configuration reference, not a compatibility test: Aider was not run against Kunavo’s endpoint, and everything above was read from Aider’s published source code and documentation. Keep a working route, run a bounded task, then check the amount recorded in your account.

Payment is made by prepaid top-up, without a subscription. The minimum top-up is $10 — a minimum for funding the balance, neither a price per task nor a plan. Checkout is handled by Stripe: cards (Visa, Mastercard, American Express), Apple Pay, Google Pay, and Link; SEPA Direct Debit is currently not offered. Pay by Bank is available only to buyers in the United Kingdom, Ireland, and Finland; in France, Stripe still limits it to a private preview (Stripe documentation, read on October 3, 2026). The balance does not expire. Details are available on the Billing page; you can create an account when you are ready. The full English version of this guide is Aider pricing.

Frequently asked questions

Aider’s repository, package, documentation, and leaderboard data verified on October 1, 2026; the behavior described is based on the published v0.86.2 source code. Kunavo rates come from the live catalog, and each numerical example is illustrative token arithmetic, not a measured task cost.