AirCode vs Codex CLI

At a glance

How the agent workflows differ

Both are terminal coding agents built mostly in Rust, and both read your project, plan, edit files, run commands and report back. They differ in what they treat as the default way of working.

AirCode: loops, checks and limits

AirCode is built for work that keeps going without you. aircode loop "make the tests pass" --until "npm test" repeats until the check passes or a limit is reached: --max-iter (10 by default), --max-cost (5.00) and --max-minutes (60). With --checkpoint git, it captures progress in Git and can revert agent-made regressions.

aircode watch --check "npm test" runs your check on every save and starts a fix iteration when it fails. Project-provided MCP servers and check commands stay off until you approve them, and /revert undoes the agent’s last edit.

Codex CLI: sandboxing and approvals

Codex puts its sandbox front and center. By default it writes only inside the workspace and has no network access; --sandbox switches between read-only, workspace-write and full access, and approval settings decide when it asks you first. On macOS it uses Seatbelt, and on Linux bwrap with seccomp.

For automation, codex exec runs in a read-only sandbox by default, emits JSONL with --json, can validate output against a schema with --output-schema, and can resume a previous run. OpenAI’s docs recommend creating Git checkpoints before and after a task so you can revert changes.

Shared ground

Both support MCP (Codex with STDIO and Streamable HTTP servers; AirCode with stdio and streamable HTTP), both read project instructions (AGENTS.md for Codex; AIRCODE.md, plus AGENTS.md, for AirCode), and both resume earlier sessions.

Models and routing

Codex CLI runs OpenAI’s models. You choose the model with /model, -m or model in the config file, and tune model_reasoning_effort. It can also run local open-weight models through Ollama or LM Studio, or a custom provider you define.

AirCode sends every request through AirLLM, where the default routing model, airllm-auto, matches each step to a suitable model across supported providers. airllm-openai keeps requests within OpenAI’s model family, and airllm-anthropic within Anthropic’s. The routing models page explains each one, and What is an LLM router? covers the idea behind them.

Platforms and availability

Codex CLI is available today on macOS, Linux and Windows, installed with a shell or PowerShell script, npm or Homebrew. Signing in with ChatGPT gives you plan access; signing in with an API key bills at standard API rates and leaves out the cloud features.

AirCode CLI is available today for macOS and Linux, with a Windows 11 x64 installer in the Developer Guide. You sign in with aircode login and choose which AirLLM API key it uses. AirCode Studio is coming soon for macOS, Windows and Linux.

When to choose each

Choose Codex CLI if…

• You want an open-source agent you can read, build and modify under Apache-2.0.
• You already have a ChatGPT plan and want Codex in the terminal, the IDE and the cloud.
• You want to run local models through Ollama or LM Studio.

Choose AirCode CLI if…

• You want each step routed to a suitable model across providers, without choosing one yourself.
• You run long, unattended jobs: loops until the tests pass and fix-on-save, with limits on iterations, time and cost.
• You want one AirLLM key and one bill for the agent and for other tools you route through Air Router.

You don’t have to choose: Codex CLI with Air Router

Codex can use Air Router as a custom provider. Current Codex releases only accept endpoints that implement the Responses API, and Air Router serves /v1/responses, so set wire_api = "responses". Provider settings belong in your user-level ~/.codex/config.toml, because a project’s own config file ignores them.

# airllm-openai stays on OpenAI models; airllm-auto routes across providers
model = "airllm-openai"
model_provider = "airllm"

[model_providers.airllm]
name = "AirLLM"
base_url = "https://api.airllm.ai/v1"
env_key = "AIRLLM_API_KEY"
wire_api = "responses"

Then export AIRLLM_API_KEY with your AirLLM key and start codex. The Codex setup guide has the steps for each operating system, and the Codex CLI router guide explains what routing changes.

About this comparison. Compiled from each product’s public documentation on October 20, 2026. Details change often, so check the sources below. Product names and trademarks belong to their owners.

Frequently asked questions

Related

Sources

• AirLLM — AirCode CLI
• AirLLM Developer Guide — AirCode CLI and AirCode Studio
• AirLLM Terms — section 10, AirCode software license
• openai/codex on GitHub (README and license)
• Codex docs — CLI
• Codex docs — Authentication
• Codex docs — Pricing
• Codex docs — Models
• Codex docs — Advanced configuration
• Codex docs — Gateway compatibility
• Codex docs — Approvals and security
• Codex docs — Non-interactive mode