Claude Code vs Codex CLI

Claude Code is Anthropic's closed-source terminal agent for Claude models. Codex CLI is OpenAI's open-source agent for OpenAI models. Both offer headless runs, MCP, subagents and a sandbox.

Where they differ

Licence. Claude Code is proprietary, and its GitHub repo holds no source. Codex CLI is Apache-2.0.
Sign-in and cost. Claude Code needs a paid Anthropic plan or an API account, and the free claude.ai plan is excluded. Codex CLI signs in with a ChatGPT plan, which has a free tier with usage limits, or uses an OpenAI API key.
Models. Claude Code runs Claude models only, and gateways to other models are unsupported. Codex CLI uses OpenAI models, and OpenAI's pages do not document local models.
Platforms. Claude Code documents git worktrees and phone access through Remote Control, but has no sandbox on native Windows. OpenAI documents a native Windows sandbox for Codex CLI.

Which fits

Pick Claude Code if you pay for Claude and want worktrees, CI integration and phone access.
Pick Codex CLI if you want open-source code, or have a ChatGPT plan and run Windows.

	Claude Code	Codex CLI
platforms	macOSWindowsLinux	macOSWindowsLinux
price	SubscriptionThe CLI needs a Claude Pro, Max, Team, Enterprise or Console (API) account; the free claude.ai plan does not include it.	FreeThe CLI itself is free (Apache-2.0). Model cost comes from your OpenAI account (seen 2026-10-28): sign in with a ChatGPT plan, where Codex is…
source	Proprietary	Open source
MCP	✓ yes	✓ yes
Multi-agent	✓ yes	✓ yes
Headless / CI	✓ yes	✓ yes
Sandbox	✓ yes	✓ yes
Local models	– no	? not verified
Your own API key	✓ yes	✓ yes
Git worktrees	✓ yes	? not verified
SSH	? not verified	? not verified

Claude Code

Codex CLI

platforms

macOSWindowsLinux

price

SubscriptionThe CLI needs a Claude Pro, Max, Team, Enterprise or Console (API) account; the free claude.ai plan does not include it.

FreeThe CLI itself is free (Apache-2.0). Model cost comes from your OpenAI account (seen 2026-10-28): sign in with a ChatGPT plan, where Codex is…

source

Proprietary

Open source

MCP

✓ yes

Multi-agent

✓ yes

Headless / CI

✓ yes

Sandbox

✓ yes

Local models

– no

? not verified

Your own API key

✓ yes

Git worktrees

✓ yes

? not verified

SSH

? not verified

Codex CLI is stronger at

Apache-2.0 with a very active repo (release published 2026-10-27)

Two sign-in routes: ChatGPT plan (including a $0 Free tier listed on the pricing page) or an API key

Documented codex exec for CI/scripts, local and remote MCP servers, subagents, and a permissions command showing the active sandbox

Native Windows sandbox documented, plus standalone installers for macOS, Linux and Windows

Claude Code is stronger at

Documented non-interactive mode (claude -p), GitHub Actions and GitLab CI integration, and an Agent SDK

Subagents, background agents and git worktrees (claude --worktree) for parallel sessions

OS-level sandboxed Bash tool on macOS (Seatbelt), Linux and WSL2 (bubblewrap)

Same engine across terminal, IDEs, desktop app, web, and phone via Remote Control and the Claude mobile app

which to pick

pick Codex CLI if

ChatGPT subscribers or OpenAI API users who want an open-source, sandboxed terminal agent.

pick Claude Code if

Developers who already pay for Claude and want a scriptable, sandboxable agent that works across terminal, IDE, CI and phone.

skip Codex CLI if

Anyone who needs to run it on non-OpenAI or local models and wants that officially documented.

skip Claude Code if

People who want an open-source tool or need to run local or non-Anthropic models.

Claude Code vs Codex CLI: what differs

Where they differ

Which fits

More

Codex CLI is stronger at

Claude Code is stronger at

which to pick