Guide · Claude Code & Codex

Claude Code vs Codex limits, side by side

Q: How do I see both at once?

Each CLI only reports its own usage: /usage in Claude Code, /status in Codex. A menu bar app like Quotchi can show both side by side.

Two tools, two allowances, two sets of windows. Here’s how they compare, line by line, from the official docs.

By Thomas Kanze

Last checked October 9, 2026. Limits change; the sources linked below are the final word.

See how Quotchi works

Quotchi standing between two laptops, one glowing coral and one glowing green

The direct answer

Both tools meter you in windows, not message counts, and both can run a five-hour window and a weekly window at the same time. Claude Code (Pro and Max) always has a five-hour session limit and a weekly limit, and the allowance is shared with claude.ai. Codex limits depend on your ChatGPT plan; OpenAI says Pro currently has no five-hour limit, and weekly limits may also apply. Check Claude with /usage and Codex with /status or its usage dashboard. The allowances are separate, so when one tool is out, the other may still have room.

Side by side

Claude Code vs Codex limits, as documented on October 9, 2026
	Claude Code	Codex
Which plans include it	Pro, Max 5x, Max 20x, Team, Enterprise. The pricing page doesn’t list Claude Code for Free.	ChatGPT Free, Go, Plus, Pro, Business, Edu and Enterprise.
Listed prices (USD)	Pro $20/month, or $17/month billed yearly. Max 5x $100/month, Max 20x $200/month (web subscriptions). Team from $25 per seat monthly.	Free $0, Go $8, Plus $20, Pro from $100 a month. Business $25 per user monthly, or $20 billed yearly. Enterprise and Edu: contact sales.
Short window	Session limit, resets every five hours.	A five-hour window on some plans. OpenAI says Pro currently has none.
Long window	Weekly limit at a fixed weekly time for your account. Max also has a separate Fable weekly limit.	Weekly limits “may also apply”. Your dashboard shows the reset.
Shared with	claude.ai and Claude Desktop, plus IDE use.	Local messages (CLI, IDE, app) and cloud chats; cloud tasks may use more.
Message counts	None published. Depends on length, features, model and effort.	Estimated ranges for some plans, explicitly “not fixed message limits”.
Check usage	`/usage`, Settings → Usage on claude.ai, status line `rate_limits`.	`/status`, chatgpt.com/codex/settings/usage, `/statusline` rate-limits item.
Keep going past a limit	Usage credits at standard API rates (`/usage-credits`); occasional one-time limit resets.	Buy credits (Plus, Pro); banked resets if your account has one; an API key for extra local chats.
Waits for the reset for you	Yes, auto-continue in interactive sessions (v2.1.234+).	An agent mid-turn can finish that turn, subject to fair use.

Prices are what Anthropic’s and OpenAI’s pricing pages listed on the day we checked. Tax, region and app-store billing can change them, so treat the Claude pricing page and Codex pricing page as the final word.

The window structure of each tool. Which windows apply to you depends on your plan; each tool’s own usage view is the authority.

How the windows work in each tool

Claude Code

Anthropic’s help center says both Pro and Max have a five-hour session limit and a weekly limit, and that these are shared across Claude and Claude Code. The session limit resets every five hours. For Max, the weekly limit applies across all models and resets at a fixed time each week assigned to your account, regardless of when you subscribed. Max also has a separate weekly limit for Fable. On top of those, you can meet an Opus or Sonnet limit that only covers one family.

Every request counts against both the session and the weekly allowance. A heavy few days can empty the weekly allowance while session windows keep resetting, and then only the weekly reset helps.

Codex

OpenAI’s Codex pricing page says your ChatGPT plan sets the allowance, publishes estimated local messages per five-hour period for Plus and Standard Business, and says weekly limits may also apply. It also says Pro plans currently have no five-hour limit. Local messages and cloud chats share the allowance, and cloud tasks may use more of it.

OpenAI’s help center describes the five-hour window as starting with your first message after the previous window ends, and the weekly period as following your previous weekly reset date. A banked or purchased reset can move that date.

Not message counts

Neither company gives you a fixed number of prompts. Anthropic says usage depends on conversation length and complexity, features, model and effort level. OpenAI says model choice, context, reasoning, tool use, retrieval and caching all affect usage, and that prompt length alone isn’t a reliable estimate. Long agentic tasks cost more in both tools.

How to check usage in each

Where the numbers live
Place	Claude Code	Codex
In the CLI	`/usage`: plan bars, reset times and a breakdown of what used it.	`/status`: remaining limits in an active session, per OpenAI’s pricing page.
In the browser	claude.ai → Settings → Usage.	chatgpt.com/codex/settings/usage.
Always on screen	Status line script reading `rate_limits.five_hour` and `seven_day`.	`/statusline` → turn on the rate-limits item.
Token history	`/usage` Session block, `/insights`.	`/usage daily`, `weekly` or `cumulative`.

# Claude Code
/usage

# Codex CLI
/status
/usage weekly

Each command only knows about its own tool. That’s the gap Quotchi fills: Claude Code’s and Codex’s windows and reset times side by side in the Mac menu bar, using each tool’s existing sign-in. More detail in how to check Codex usage and Claude Code usage limits.

How resets and overages differ

Waiting. Claude Code can wait in the open session and continue the task after the reset. Codex lets an agent that hits your limit mid-turn finish that turn, subject to fair use.
Paying. Claude’s usage credits are billed at standard API rates and you set a monthly cap. Codex Plus and Pro users can buy credits; OpenAI says included usage is used first and prices depend on your plan.
One-time resets. Anthropic occasionally offers eligible plans a free limit reset that restores the session or the weekly limit, without moving your weekly reset day. OpenAI’s banked resets refresh both the five-hour and weekly windows and do change your weekly reset date.
API keys. In Claude Code an API key in your environment silently replaces your subscription. In Codex, OpenAI positions an API key as a deliberate way to run extra local chats at API rates.
Support. Neither offers resets on request. OpenAI’s help center says support doesn’t reset Codex limits; Anthropic’s resets are offers you use yourself.

Worked example

Tuesday, 3:30pm. Claude Code stops: You've hit your session limit · resets 6:05pm. Your Codex dashboard shows plenty of room in its current window. You commit, write a three-line handoff note, and let Codex carry on with the tests while Claude Code waits for 6:05pm to continue the refactor. Neither allowance was touched by the other.

Running both: a practical setup

Plenty of developers keep one plan from each company. If that’s you, a little setup makes the second allowance a real safety net instead of a scramble:

Share instructions between tools. Claude Code reads an AGENTS.md file as project instructions when there’s no CLAUDE.md (v2.1.277 or later), so one file can brief both agents on your conventions.
Keep work committable. Small commits make a mid-task switch painless. The other agent picks up from the repository, not from a conversation it never saw.
Put both numbers on screen. Claude Code’s status line can show its five-hour and weekly percentages; Codex’s /statusline has a rate-limits item. Each line shows only its own tool.
Know each tool’s “keep going” option. Usage credits for Claude (/usage-credits), credits or a banked reset for Codex. Decide in advance what you’re willing to pay for.
Don’t let API keys surprise you. Check /status in Claude Code for a stray ANTHROPIC_API_KEY; in Codex, an API key is billed under API pricing.

Splitting by task type works well too: one tool for long refactors that benefit from auto-continue, the other for parallel cloud tasks. The allowances never interact, so the split is entirely up to you.

When to use which

We’re not going to rank the models; that changes with every release. Limits give you a more practical rule:

Already pay for one? Start there. Both $20 tiers include their coding agent.
Hit the short window often? Look at plans where it’s larger or absent. Max multiplies Claude’s per-session allowance; OpenAI says Codex Pro currently has no five-hour limit.
Live in claude.ai or ChatGPT all day? Remember the sharing. Claude Code shares limits with claude.ai; Codex shares with ChatGPT’s agentic features once their pricing applies.
Run long unattended work? Claude Code’s auto-continue is built for waiting out a session limit. Codex cloud tasks can run without your laptop but may use more allowance.
Want insurance? Two $20 plans give you two independent allowances. When one is out, switch.

Not sure which kind of builder you are? Our 60-second quiz maps your habits to a plan suggestion.

Running both? Quotchi shows Claude Code and Codex usage, reset times and which tool has room, in one glance from the Mac menu bar. It can also read Gemini usage from the Gemini Mac app if you turn that on. Free and local-first.

See how Quotchi works

Common mistakes and myths

“Codex and Claude Code have the same limits because they cost the same.”: The $20 tiers line up on price, but the windows, sharing rules and overage options differ.
“Codex Pro has no limits.”: OpenAI says it has no five-hour limit currently. It still has an included allowance, and weekly limits may apply.
“The published Codex message ranges are my quota.”: OpenAI calls them estimates, not fixed limits, and they change with models.
“Using claude.ai doesn’t affect Claude Code.”: It does. Anthropic says all activity in both counts against the same limits.

What changed recently

As of October 9, 2026: OpenAI’s Codex pricing page says GPT-5.5 retires from ChatGPT, ChatGPT Work and Codex on all plans on October 14, 2026, and that Fast and Ultrafast modes use included limits and credits at different rates. On the Claude side, Claude Code v2.1.234 added automatic continue after a usage limit, and Max plans list a separate Fable weekly limit.

Sources

Official pages we read on October 9, 2026. If anything here disagrees with them, they win.

FAQ

Is Codex cheaper than Claude Code?

At the entry level they’re listed at the same price: Claude Pro and ChatGPT Plus are both $20 a month. ChatGPT Free and Go also include Codex. Higher tiers differ, so compare the current pricing pages.

Do Claude Code and Codex share a limit?

No. They’re separate companies and separate allowances. Claude Code shares its limit with claude.ai; Codex shares its allowance between local and cloud use.

Which one has a five-hour limit?

Claude Code on Pro and Max always has a five-hour session limit. Codex has a five-hour window on some plans; OpenAI says Pro currently doesn’t.

How do I see both at once?

Each CLI only reports its own usage: /usage in Claude Code, /status in Codex. A menu bar app like Quotchi can show both side by side.

Can I buy more usage in both?

Yes. Claude Pro and Max can turn on usage credits billed at API rates. Codex Plus and Pro users can buy credits after reaching their limit.

Do weekly limits reset at the same time?

No. Claude’s weekly reset is a fixed time assigned to your account. Codex’s follows your previous weekly reset date and can move if you use a banked or purchased reset.

Does an API key help?

In Codex, OpenAI says anyone can run extra local chats with an API key at standard API rates. In Claude Code, an API key in your environment overrides your subscription, so set it only on purpose.