Claude Code vs Codex: terminal coding agents compared

Claude Code and Codex CLI do the same job: an agent in your terminal that reads a repository, edits files and runs commands. The practical differences are licensing, which model family you are tied to, how you pay, and how far each reaches beyond the terminal. Checked on 2026-10-30 against Claude Code 2.1.314 (released October 30) and Codex CLI 0.166.0 (October 29).

Quick comparison

Ranks in the item list are alphabetical, not a verdict.

Verdict

Pick Codex CLI if you want open source code you can audit or patch, or if your team already pays for ChatGPT. Pick Claude Code if you want one agent across terminal, JetBrains, desktop and browser, or if you need scheduled cloud runs and coordinated parallel agents. If you already have a strong preference for Claude or GPT models, that settles it: each tool is built around its maker's models.

Where each one runs

Claude Code's docs describe four surfaces: the terminal CLI, IDE extensions for VS Code and JetBrains, a desktop app and the web. The web version needs no local setup, runs long tasks and parallel sessions, and is also in the Claude mobile apps. The desktop app includes Claude Code, so you do not need the CLI separately, but it needs a paid plan.

The Codex README centres on the CLI and points elsewhere for the rest: an IDE extension for VS Code, Cursor and Windsurf, a desktop app started with codex app, and the cloud agent Codex Web at chatgpt.com/codex. All of them draw on the same ChatGPT plan allowance when you sign in with ChatGPT.

Plans, API keys and limits

Claude Code. The free Claude plan does not include Claude Code; Pro does, at $17 per month billed annually or $20 billed monthly. Max starts from $100 per month with 5x or 20x Pro usage. Team standard seats are $20 per seat per month billed annually, $25 monthly. Without a plan, you can use an Anthropic Console account, and the docs say the CLI, VS Code and JetBrains also accept third-party providers.

Codex. ChatGPT Plus at $20 per month includes Codex on the web, in the CLI, the IDE extension and on iOS. Pro plans are $100, $200 or $500 per month. The Free and Go plans get GPT-6 Luna in the desktop app only, subject to rollout. With an API key, the CLI, SDK and IDE extension bill at API rates, but cloud features such as GitHub code review and Slack are not available.

On limits, OpenAI publishes estimated local messages per five-hour window per model for Plus and Business, adds that weekly limits may apply, and says Pro currently has no five-hour limit. Anthropic describes plan usage only relative to Pro. Neither publishes a fixed task count, so compare on your own workload before committing to a higher tier.

Project instructions and extensions

Codex reads AGENTS.md for project guidance. Claude Code reads CLAUDE.md at the start of each session and can read an existing AGENTS.md instead, which lets a repository serve both tools from one file. Both connect to MCP servers. Claude Code adds skills, hooks and plugins; the Codex docs list skills, hooks and subagents.

Automation and CI

Claude Code runs headless with claude -p, for example piping a log into it, and the docs list GitHub Actions and GitLab CI/CD for code review and issue triage. Cloud scheduled tasks run on Anthropic's side even when your computer is off and can trigger on API calls or GitHub events.

Codex has codex exec for non-interactive runs, a Codex GitHub Action and a Codex SDK for controlling local agents from code. OpenAI's pricing page recommends API key sign-in for shared environments such as CI.

Release pace

Both ship often. Claude Code had releases on October 26, 27 and 30. Codex shipped stable 0.165.0 on October 27 and 0.166.0 on October 29, with alpha builds on npm in between; see Codex CLI 0.166.0 for what the latest one changed. Pin versions in CI for either tool.

Who should not use either

If you need an agent that runs on any model, including local ones, neither is the best fit: look at OpenCode or Aider. If you want a fully managed agent that works in its own cloud machines, see Devin vs Claude Code.

How we compared

We read the anthropics/claude-code and openai/codex READMEs and GitHub metadata, the Claude Code docs overview, the Claude pricing page and the Codex pricing page on 2026-10-30. Stars are rounded GitHub API counts. We did not benchmark the two agents on the same tasks, so nothing here ranks their code quality.

Limitations

Prices are as listed on the vendors' pages on 2026-10-30 in US dollars and change often. Model names and limits per plan were not compared in depth, because both vendors change them faster than this page can track.

Related: Claude Code hub, Codex CLI hub, coding agents category, Aider vs Claude Code, Qwen Code vs Cline and OpenCode vs Cline vs Aider.