Codex vs Claude Code: which should builders use?
Both are agentic coding tools that read your repo, edit files and run commands. OpenAI Codex works from an open-source (Apache-2.0) CLI, IDE extensions, the ChatGPT desktop app and Codex cloud, and comes with ChatGPT plans. Anthropic's Claude Code works from the terminal, VS Code and JetBrains, a desktop app and the web, and comes with paid Claude plans.
| Made by | OpenAI | Anthropic |
|---|---|---|
| Where it runs | CLI; IDE extension for VS Code, Cursor and Windsurf (plus Xcode and JetBrains integrations); ChatGPT desktop app; Codex on the web; iOS | Terminal CLI; VS Code extension (also installs in Cursor); JetBrains plugin; desktop app; web at claude.ai/code; Claude iOS and Android apps |
| Cloud and team features | Codex cloud tasks, @codex on GitHub issues and PRs, automatic PR review, Slack and Linear integrations (ChatGPT plans, not API key) | Cloud sessions on the web, scheduled Routines, @Claude in Slack, GitHub Actions and GitLab CI/CD, GitHub Code Review |
| Free access | Limited: ChatGPT Free and Go ($8/mo) get Codex in the desktop app on GPT-6 Luna, subject to rollout | None: Claude Code isn't included in the Free plan |
| Individual plans | Plus $20/mo (web, CLI, IDE extension, iOS, cloud integrations); Pro $100, $200 or $500/mo | Pro $20/mo ($17/mo billed annually); Max from $100/mo with 5x or 20x Pro usage |
| Business plans | Business $20/user/mo billed annually (2+ users), $25 billed monthly; Enterprise and Edu through sales | Team: Standard seats $25/mo and Premium seats $125/mo billed monthly; Enterprise $20/seat/mo plus usage at API rates |
| Usage limits | Allowance per 5-hour window on Plus and Business (OpenAI estimates 15-160 local messages on GPT-6.1 Sol); weekly limits may apply; Pro currently has no 5-hour limit; buy ChatGPT credits to keep going | Rolling 5-hour session window plus weekly limits on paid plans; one pool shared with Claude chat; turn on usage credits at API rates when you hit a limit |
| Pay as you go | Sign in with an API key: CLI, SDK and IDE extension billed at API rates, no cloud features | Anthropic Console / API key at API rates; the CLI, VS Code and JetBrains also work through Amazon Bedrock, Google Cloud and Microsoft Foundry |
| Models | GPT-6.1 Sol (recommended for complex work), GPT-6 Astra, GPT-6 Sol, GPT-6 Luna | Opus 5.5 by default on paid plans; Sonnet 5.5 and Haiku 5.5; Fable 5.1 by choice (usage credits on Pro, up to 50% of weekly limits on Max) |
| Open source | Yes: the CLI is Apache-2.0 (github.com/openai/codex) | No: proprietary, © Anthropic PBC, all rights reserved |
| Project instructions file | AGENTS.md | CLAUDE.md (can also read AGENTS.md) |
| Pricing (approx.) | Limited on ChatGPT Free and Go ($8/mo); Plus $20/mo; Pro from $100/mo; or API usage | Included with Claude Pro ($20/mo US) and Max plans, or pay via API usage |
| Our rankings | not in a ranking yet | #9 in Vibe coding, #2 in Coding agents |
How the usage limits work
Neither tool has a fixed message count. Codex shares one allowance with ChatGPT Work, and OpenAI publishes estimated local messages per 5-hour window by model (on Plus, roughly 15-160 for GPT-6.1 Sol and 350-3,000 for GPT-6 Luna). Cloud tasks can use more of the allowance than local ones. Claude Code draws from the same pool as Claude chat on web, desktop and mobile. That pool resets on a rolling 5-hour window, and paid plans add weekly limits. Both let you keep going past the limit: Codex with ChatGPT credits or an API key, and Claude Code with usage credits at API rates or a Console account.
Which is cheaper?
At the entry level, they cost the same: $20/mo for ChatGPT Plus or Claude Pro (Claude Pro is $17/mo billed annually). Codex has a limited free tier through ChatGPT Free and Go, and Claude Code does not. For heavy use, ChatGPT Pro starts at $100/mo and Claude Max starts at $100/mo.
Pick OpenAI Codex if…
You already pay for ChatGPT, want an open-source CLI you can read and fork, or want cloud tasks and automatic GitHub PR review included in the $20 Plus plan. It's also the one you can try on a free plan.
Pick Claude Code if…
You already pay for Claude, want the same agent across terminal, IDE, desktop and browser with CLAUDE.md, skills and hooks, or need to run it through Amazon Bedrock, Google Cloud or Microsoft Foundry.
Our verdict
Both start at $20/mo and both share a rolling 5-hour usage pool with their chat app, so price won't decide it. Pick Codex if you want an open-source CLI, a free way to try it, and GitHub PR review in ChatGPT Plus. Pick Claude Code if you're already on Claude or need it on your cloud provider. Either one is a strong terminal agent next to an editor like Cursor.
Facts and prices checked on vendor pages on Oct 14, 2026: Codex pricing, Codex models, ChatGPT pricing, openai/codex on GitHub, Claude pricing, Claude Code overview, Claude Code model config. Plans and limits change often, so confirm before you buy.