Skip to main content
Using Claude

Claude Code vs Codex: Anthropic's CLI vs OpenAI's Coding Agent (2026)

Comparing Claude Code (Anthropic) and Codex (OpenAI) — supervised local terminal vs async cloud delegation, multi-interface, open source, models, pricing, and which to pick.

ByUpdated
🌐 This article was machine-translated and may contain inaccuracies. Read the Korean original if in doubt.

"Codex" is OpenAI's coding agent (relaunched in 2025 — not the deprecated 2021 Codex model). It's often compared with Claude Code (Anthropic), and the core difference is one thing — Claude Code is a supervised local terminal workflow, while Codex is a multi-interface tool that adds asynchronous cloud delegation on top of a local CLI.

🟢 Model references match the current lineup · model notice · Fable subscription
🟢 Model references match the current lineup · Claude Opus 5.5 / Claude Sonnet 5 / Claude Haiku 4.5 (higher tier: Claude Fable 5.1). This notice changes only when Anthropic ships a new model.

Fable 5 and 5.1 subscription (updated September 7, 2026): Claude Fable 5.1, released September 1, 2026, is the current Fable model and Fable 5 is now legacy. Plan terms are the same for both — Max and Team Premium plans include Fable at up to 50% of the weekly usage limit; Pro and Team Standard use usage credits (

🟢 Model references match the current lineup · Claude Opus 5.5 / Claude Sonnet 5 / Claude Haiku 4.5 (higher tier: Claude Fable 5.1). This notice changes only when Anthropic ships a new model.
0/M input, $50/M output tokens). The one-time
🟢 Model references match the current lineup · Claude Opus 5.5 / Claude Sonnet 5 / Claude Haiku 4.5 (higher tier: Claude Fable 5.1). This notice changes only when Anthropic ships a new model.
00 credit applied only to the Fable 5 transition and is not offered for 5.1. Some coding and debugging requests may be answered by an Opus model due to a security classifier (both models). See the Fable 5.1 guide and the Fable 5 availability guide for details.

Claude Code vs Codex at a glance Item Claude Code Codex (OpenAI) Made by Anthropic OpenAI Form Terminal CLI, supervised CLI + cloud + IDE Models Claude (Opus, Sonnet) GPT-5 family Strength Local supervised, multi-file Async cloud delegation CLI open source No Yes (Apache-2.0) Pricing Included in $20 Pro/Max Included in ChatGPT plans Both handle daily work on a ~$20 plan. Benchmark scores vary by source,version, and date, so they're hard to state as fact. As of July 2026.

What is Codex?

OpenAI Codex comes in three forms:

  • Codex CLI — a terminal coding agent. Open source (Apache-2.0).
  • Codex cloud — hand it a task in ChatGPT and an isolated cloud sandbox clones your repo, edits code, runs tests, and presents changes (or a PR). You can delegate and do other work meanwhile.
  • IDE extension — integrated into your editor.

It runs on GPT-5-family models and is included in ChatGPT plans (Free through Pro).

What is Claude Code?

Anthropic's Claude-only terminal CLI agent. It reads and edits files locally but defaults to a supervised flow that asks for approval before writes and commands. 1M context on Opus, with programmable hooks and subagents for fine-grained control. Included in Pro/Max. (See how to use Claude Code.)

Key differences

  • Workflow: Claude Code is a supervised local flow with the developer in the loop. Codex also runs a local CLI but shines at async cloud delegation (give a spec, get a result).
  • Isolation vs control: Codex uses OS-kernel sandboxing for strong boundaries but coarser control; Claude Code uses app-level hooks — weaker boundaries but finer control. → Kernel sandboxing for untrusted external code; hooks for enforcing org coding standards.
  • Ecosystem: already on ChatGPT → Codex; already on Claude → Claude Code.
Both are terminal-based coding agentsSimilar features — what decides itModel & ecosystemWhich vendor\u2019s modeland tools fit youSubscription & priceIncluded in a planyou already have?Personal tasteCommands, UX, flowthat fit your handsStrengths vary by task/version — try both and pick · as of 2026.6

Which to choose

  • Claude Code — when you want to review each step locally across multi-file, large codebases.
  • Codex — when you throw detailed specs and delegate asynchronously, or do lots of terminal-native work.
  • Both — each is enough for daily work on a ~$20 plan, so many developers run both (~$40/mo combined) and cross-check answers on hard problems.

Curious about Cursor (a GUI editor with multiple models)? See Claude Code vs Cursor.

Disclaimer: The base model versions, interfaces, context, pricing, and benchmark scores for both tools change often. Commonly cited benchmark numbers vary by source, version, and date, so this article doesn't state specific scores as fact (performance is a general tendency). Verify the latest in each official doc (code.claude.com, OpenAI Codex docs). This site is not affiliated with Anthropic or OpenAI.

Side-by-Side Decision Table

Use this table when you have a specific project type in mind. Every row maps to a real workflow difference — not just a spec sheet.

Situation Pick Claude Code Pick Codex
You want to watch every file edit live ✅ Supervised terminal — you approve each step ❌ Cloud tasks run without you
You want to queue tasks and come back later ❌ Requires active session ✅ Async cloud delegation built-in
Your codebase is sensitive / offline ✅ Runs entirely local ⚠️ Cloud mode sends code to OpenAI servers
You already use VS Code or JetBrains daily ⚠️ Terminal-first; IDE integration is secondary ✅ Native IDE plugins available
You prefer open-source tooling ✅ Claude Code is open source ❌ Closed source
Budget: pay-per-use vs flat subscription Included with Claude Pro / Max — check claude.ai/pricing for current plan limits Usage-based cloud tasks — check platform.openai.com/pricing for current rates

Quick-Start: Try Each Tool in Under 5 Minutes

Prerequisites: a terminal open, the relevant API access, and Node.js only if you install via npm (the Claude Code npm package requires Node.js 22+ since v2.1.198; the native installer needs no Node.js — official docs, September 2026; check the Codex CLI README for its own requirement) (Claude Pro/Max for Claude Code; OpenAI account with Codex enabled for Codex).

Claude Code — 3 steps

  1. Install globally (verify package name at code.claude.com before running):
    npm install -g @anthropic-ai/claude-code
  2. Open your project folder in a terminal, then run:
    claude
  3. Success signal: You see a prompt like Claude Code > and Claude immediately lists the files it can see. Type a plain-English task — e.g. "Add input validation to login.js" — and approve each proposed edit.

Codex CLI — 3 steps

  1. Install (verify at platform.openai.com/docs/codex before running):
    npm install -g @openai/codex
  2. Run in your project folder:
    codex
  3. Success signal: The CLI confirms your OpenAI key and shows a task prompt. For cloud async tasks, check the Tasks tab in your OpenAI dashboard — the agent runs in a sandboxed environment and posts results when done.

Stuck? The most common blocker is a missing or expired API key. For Claude Code, set ANTHROPIC_API_KEY in your shell environment. For Codex, set OPENAI_API_KEY. Both tools will print a clear error message pointing to the missing variable.

Was this helpful?

Keep reading

Running Claude Code and Codex Together: A Role-Split Workflow

Some developers don't pick one over the other — they run Claude Code and Codex side by side, splitting implementation and review between them. Here's how to set that up and what to watch for.

90

Anthropic Updates Roundup - 2026/09 (3 items)

A monthly roundup of 3 Anthropic updates in 2026/09

10

Claude Opus 5.5 Complete Guide: Fable 5.1-Level Performance for Less Than Opus 5 — Pricing, Plans, Migration

Released September 22, 2026, Claude Opus 5.5 performs at Fable 5.1 level on most work and costs 40% less than Opus 5. API pricing ($4/$20), plan availability, higher five-hour limits and the usage reset, plus the five breaking changes when migrating from Opus 5 — all from official docs.

75

Claude's Skills Explore Tab Explained — What 60 Plugins Do, How to Use Them, and Whether They Run Automatically (Sept 2026)

What each plugin listed under Settings → Skills → Explore does, how you call it, and whether it can run with nobody around — based on the SKILL.md files of 60 installed plugins and Anthropic's official docs.

65