# Airuncode

Airuncode runs coding-agent workflows on your computer with your own model keys or local GGUF models. Paid plans add multi-agent work and context snapshots without a token markup.

Canonical URL: https://aiidelist.com/ide/airuncode

Language: en

Updated: 2026-09-11

## Overview

- Category: Terminal / CLI Coding Agents
- A local coding-agent runtime with provider keys, local models, and optional multi-agent execution.
- Editor base: Standalone
- Platforms: macOS, Windows, Linux
- Local model support: Yes
- Bring your own API key: Yes

## Quick verdict

Worth a trial when model choice and direct provider billing matter; compare the free activation limits with your working schedule.

## Best for

- Developers choosing their own model providers
- Users evaluating local GGUF inference

## Strengths

- Offers local models alongside external providers.
- Separates software subscription charges from provider inference costs.

## Limitations

- Users expecting unlimited cloud-model use on the free software tier
- The free tier limits provider-key activation; unlimited BYOK requires Pro.
- Cloud model requests still leave the machine even though the runtime is local.

## Choose the model path first

Airuncode's most useful distinction is the choice between local inference and your own cloud provider. These configurations have different hardware, cost, and data-handling implications. A local GGUF model needs enough memory on your computer; a cloud provider needs an account and sends requests across the network.

## Try a fix with a measurable outcome

1. Download the build for your operating system from the [official site](https://airuncode.com/).
2. Configure a provider key or a compatible local model, then open a test repository.
3. Ask for one small change and identify the relevant test command in the task.
4. Review the generated diff, test output, and recorded cost before enabling a broader multi-agent workflow.

For example, use an existing failing parser test as the acceptance condition. If the agent proposes changing the test instead of fixing the parser, review that decision before continuing.

## Compare total cost and supervision

Zero token markup describes inference billing, not the price of every software feature. Compare the plan cost plus provider usage for the tasks you actually run. Test-repair automation is useful feedback, but a passing suite alone does not establish that a change meets the original requirement.

## Features

### Model access

- Bring your own provider keys
- Local GGUF models through llama.cpp
- Provider cost tracking without an Airuncode token markup

### Task execution

- Multi-agent workflow on paid plans
- Context snapshots
- Test feedback for supported runners including Vitest, Jest, pytest, and Cargo

## Pricing

freemium plus model-provider usage

- Free: $0 — Two 48-hour key activations per week, local models, cost tracking, and basic V-CORE.
- Pro: $15 — month — Unlimited BYOK use, multi-agent swarm, context snapshots, and full V-CORE. Model-provider charges are separate.
- Studio: $49 — month — Up to five seats, multiple workspaces, audit export, and private endpoints. Model usage is separate.

Pricing checked: 2026-09-12

## Privacy and data handling

Execution is local. Cloud providers receive model requests when selected; local inference requires a local-model configuration.

## Alternatives

- OpenCode
- Aider

## Sources

- [Official website](https://airuncode.com/)
- [Installation](https://airuncode.com/)
- [Download](https://airuncode.com/)
- [Pricing](https://airuncode.com/)
- [Airuncode features, downloads, and pricing](https://airuncode.com/)

Last checked: 2026-09-12