AI Coding & Vibe Coding Published 2026-10-28

Claude Code vs Codex (2026): Plans, Usage Limits, and Which to Pick

How we checked: every price, limit and rule below comes from Anthropic's and OpenAI's own pages: Claude's pricing page and Help Center, and OpenAI's Codex pricing page, ChatGPT pricing page, and Help Center, all checked on October 28, 2026. Sources are listed at the end. We have no affiliate relationship with either company. Both companies change limits often and describe them as estimates, so check your own usage dashboard before you upgrade.

The short answer

  • Both start at $20 a month, and neither is sold on its own. Claude Code comes with a Claude plan (Pro, Max, Team or Enterprise). Codex comes with a ChatGPT plan (Plus, Pro, Business, Edu or Enterprise). You also get the full chat assistant from the same company.
  • Want to try before paying? Codex has limited access on ChatGPT Free and Go (GPT-6 Luna in the desktop app, subject to rollout). Claude Code isn't included on Claude's Free plan at all.
  • Want to know roughly how much you get? OpenAI publishes estimated message ranges per 5 hours for Plus (for example, 15 to 160 local messages with GPT-6.1 Sol). Anthropic publishes only multiples (Pro is at least 5x Free, Max is 5x or 20x Pro) and no message counts.
  • Work in long bursts? OpenAI says ChatGPT Pro plans currently have no 5-hour limit for Codex, though OpenAI notes that "weekly limits may also apply". Claude Pro and Max both keep a 5-hour session limit plus a weekly limit.
  • Want an annual discount? Only Claude offers one at the entry level: Pro is $17/mo billed annually ($200 up front). ChatGPT Plus and Pro, and Claude Max, are monthly only.

At a glance

Claude Code OpenAI Codex
Comes with Claude Pro, Max, Team, Enterprise ChatGPT Free and Go (limited), Plus, Pro, Business, Edu, Enterprise
Free access No (Free plan excludes Claude Code) Limited: GPT-6 Luna in the desktop app, subject to rollout
Entry paid plan Pro: $20/mo, or $17/mo billed annually Plus: $20/mo (monthly)
Higher plans Max 5x and Max 20x, from $100/mo (monthly only) Pro at $100, $200 or $500/mo (monthly only)
Published allowance Multiples only (Max is 5x or 20x Pro per 5-hour session) Estimated local messages per 5 hours, by model (Plus and Standard Business)
Limit windows 5-hour session + weekly on Pro and Max; Max adds a separate weekly limit for Fable models 5-hour on Plus; Pro currently none; weekly limits "may also apply"
Shared with Claude on web, desktop and mobile ChatGPT Work
When you run out Usage credits at standard API rates, upgrade, or wait Buy ChatGPT credits (Plus and Pro), upgrade, or wait
Pay-per-token option Sign in with a Claude Console (API) account Use an API key in the CLI, SDK or IDE extension
Check usage Settings > Usage, or /status in Claude Code Usage dashboard, or /status in the Codex CLI

How Claude Code's plans work

Claude Code runs in three places: the Code tab in Claude Desktop (Anthropic's recommended start, with nothing else to install), your terminal, and supported IDEs, including VS Code, Cursor and other VS Code forks, and JetBrains IDEs such as IntelliJ and PyCharm. All three use the same login and the same usage pool.

  • Pro ($20/mo, or $17/mo billed annually): includes Claude Code, Opus, Sonnet and Haiku models, and at least 5x the Free plan's usage per 5-hour session. Access to the Fable models is through usage credits only.
  • Max 5x and Max 20x (from $100/mo, billed monthly): 5x or 20x Pro's per-session allowance, higher output limits, priority access at busy times, and early access to new features. Fable can use up to 50% of your weekly limits. Max also includes $100 or $200 a month in API credits for building on the Claude Platform, but those API credits can't be used in Claude Code or the Claude apps.
  • Team (2 to 150 seats) and Enterprise: Claude Code is included on both. Enterprise is $20 per seat per month plus usage at API rates, billed annually. We didn't capture a Team seat price, so check Claude's pricing page.

How the limits work. Usage resets on a rolling 5-hour session, and paid plans add a weekly limit on top. On Max, the weekly limit resets at a fixed day and time assigned to your account. Your chats in Claude and your Claude Code sessions draw from the same pool, so a long research chat leaves less for coding. Anthropic doesn't publish a message count, because usage depends on conversation length, model, features and effort level.

When you hit a limit. You can wait, upgrade, or turn on usage credits, which keep you working at standard API rates until your monthly spend limit or balance runs out. Anthropic is also rolling out a small "wrapping up" allowance: if you hit the 5-hour limit mid-response, Claude Code may finish the step in progress. On Pro this applies up to once per weekly period, and on Max each time you hit a session limit.

A billing gotcha. If an ANTHROPIC_API_KEY environment variable is set on your machine, Claude Code uses that key instead of your subscription, and you're billed at API rates. Anthropic's Help Center flags this explicitly.

How Codex's plans work

Codex runs on the web, in the CLI, in an IDE extension, in the ChatGPT desktop app, and on iOS. Paid plans also get cloud features: automatic GitHub code review, delegating issues and PRs with @codex, and Slack and Linear integrations.

  • Free ($0) and Go ($8/mo): "Limited" Codex, which OpenAI's Codex pricing page describes as GPT-6 Luna at Standard speed in the desktop app, subject to rollout. Image generation isn't available on Free.
  • Plus ($20/mo): full Codex surfaces and cloud integrations, GPT-6.1 Sol, GPT-6.1 Terra and GPT-6 Luna, and the option to extend usage with ChatGPT credits.
  • Pro ($100, $200 or $500/mo, monthly only): "expanded" Codex and longer sessions. OpenAI says Pro 200 includes more usage than Pro 100, and Pro 500 the most. Only Pro 500 includes Ultrafast mode. OpenAI doesn't publish exact Pro allowances.
  • Business: 2+ users and billed annually by default, or $25 per user per month billed monthly, with SSO, MFA, and no training on business data by default.

Estimated local messages per 5 hours (Plus and Standard Business):

Model Plus
GPT-6 Astra 5 to 45
GPT-6.1 Sol 15 to 160
GPT-6.1 Terra 35 to 285
GPT-6 Sol 15 to 150
GPT-6 Luna 350 to 3,000

OpenAI stresses these are estimates, not fixed limits. Cloud tasks may use more of your allowance than local messages, and weekly limits may also apply. The wide ranges reflect how much context, reasoning and tool use each task needs. Picking a smaller model (Terra or Luna instead of Astra) is the simplest way to stretch a Plus plan.

Things that burn usage faster:

  • Fast mode uses your included allowance at 2.5x the Standard rate.
  • Ultrafast (Pro 500 only, for GPT-6 Astra and GPT-6.1 Sol) uses it at 8x.
  • Image generation uses included limits about 3 to 5x faster on average.
  • Every MCP server you connect adds context to each message. OpenAI suggests disabling the ones you don't need.

When you hit a limit. If you run out mid-turn, Codex lets the agent finish that turn, subject to fair use. After that, Plus and Pro users can buy ChatGPT credits, or anyone can run extra local chats with an API key at standard API rates.

If you're on Pro 200, check your email this week. OpenAI lowered the included allowance for new Pro 200 subscriptions (the price stays $200). Subscribers who were active between September 22 and September 29, 2026 keep the previous allowance only through October 29, 2026, then move to the lower one.

Which one should you pick?

  • Pick Claude Code if you already use Claude for writing or research and want one subscription for both, if you work mostly in the terminal or a JetBrains IDE, or if you'd like the annual Pro price ($17/mo). Keep in mind that heavy chatting eats into your coding budget.
  • Pick Codex if you want to try it free first, want published estimates of what a plan covers, or want cloud features like automatic GitHub PR reviews and Slack and Linear hand-offs included with Plus.
  • For long, bursty sessions, Codex on a Pro plan currently has no 5-hour cap, while Claude Max raises the size of each 5-hour session (5x or 20x Pro) but keeps the window.
  • For automation in CI, both support pay-per-token use: an OpenAI API key for Codex (no cloud features), or a Claude Console account for Claude Code.

If you'd rather have an AI editor than a terminal agent, see the Cursor directory entry, and for chat-to-app builders see Lovable vs Bolt vs v0. For GitHub's own assistant, see GitHub Copilot pricing. To compare the chat side of these subscriptions, see ChatGPT vs Claude vs Perplexity pricing.

FAQ

Is Claude Code free? No. Claude Code isn't included on Claude's Free plan. It comes with Pro ($20/mo, or $17/mo billed annually) and every higher plan, or you can pay per token through a Claude Console account.

Is Codex free? Partly. ChatGPT Free and Go include limited Codex access, which OpenAI describes as GPT-6 Luna in the desktop app, subject to rollout. Full Codex, including the CLI, IDE extension and cloud features, starts with Plus at $20/mo.

Do Claude Code limits reset every 5 hours? Yes. Pro and Max have a session limit that resets every five hours, plus a weekly limit. Max also has a separate weekly limit for Fable models.

Does Codex have a weekly limit? OpenAI says weekly limits "may also apply" and points you to your usage dashboard for current limits and reset times. It doesn't publish a weekly number.

Can I use the $100 or $200 API credits from Claude Max in Claude Code? No. Anthropic's Help Center says the monthly API credits included with Max can't be used in Claude Code or the Claude apps, including for usage credits.

Sources

Tools in this article

Get new AI tools in your inbox

A short digest of verified new tools and price changes. No spam.