Claude Code and Codex usage limits, explained
How Claude Code usage limits work on Pro and Max (5-hour session, weekly limit), how Codex limits work on ChatGPT plans, and how to check what’s left.
Updated
On a Claude Pro or Max plan, Claude Code has two limits: a session limit that resets every five hours and a weekly limit, both shared with claude.ai. Codex on a ChatGPT plan works the same way: a five-hour allowance (Pro plans currently have none) and weekly limits that may apply. Check what’s left with /usage in Claude Code or /status in the Codex CLI.
Everything below is as of October 2026, from Anthropic’s and OpenAI’s own help pages. Both companies change these rules often, so the official pages win if they disagree.
How Claude Code limits work
Claude Code on a subscription draws from the same allowance as the Claude apps. Everything you do in claude.ai, Claude Desktop and Claude Code counts against the same limits.
| Plan | Price | Allowance | Fable models |
|---|---|---|---|
| Pro | $20/month ($17/month billed yearly) | Base allowance | Not included: pay with usage credits |
| Max 5x | $100/month | 5x Pro per session | Up to 50% of your weekly limit |
| Max 20x | $200/month | 20x Pro per session | Up to 50% of your weekly limit |
The session limit resets every five hours. If you hit it mid-answer, you may see “Usage limit reached · wrapping up”: a small, capped allowance so Claude can finish, which still counts toward your week.
The weekly limit applies across all models. On Max it resets at a fixed time each week assigned to your account, not on a rolling 7 days from your first prompt. Your reset time shows at claude.ai/settings/usage.
Model limits. Claude Code can also say “You’ve hit your Opus limit” or “You’ve hit your Sonnet limit”. Those are per model family, so switching models with /model keeps you working. After a session or weekly limit, switching models doesn’t help.
What uses it up. No message count is published. Usage depends on the model, your effort level, how long the conversation is and the features you use. In Claude Code the big costs are long sessions (the whole conversation is re-sent with every request), cache misses after a break, subagents, and several agents at once. Agent teams use about 7x the tokens of a normal session when teammates run in plan mode.
How to check your Claude usage
/usagein Claude Code. Shows your plan’s usage bars and reset times, activity stats, and on Pro and Max a breakdown of what your usage went to: skills, subagents, plugins and MCP servers, with recommendations. Pressdorwfor the last day or week./costand/statsare aliases, and it runs even while Claude is working.- claude.ai/settings/usage. The same session and weekly bars, plus usage credits and your reset times.
/statusshows your version, model and account, not your limits. Older guides point to it for usage;/usageis the one now.
The dollar figure in /usage is an estimate at API list prices. On a subscription it isn’t what you pay.
How Codex limits work
Codex comes with every ChatGPT plan, and local CLI messages and cloud tasks share one allowance.
| Plan | Price | Codex limits |
|---|---|---|
| Free | $0 | Quick coding tasks |
| Go | $8/month | Light coding tasks |
| Plus | $20/month | Five-hour allowance, “a few focused coding sessions each week”, credits extend it |
| Pro | $100, $200 or $500/month | No five-hour limit right now, included allowance still applies |
| Business | $25/user/month ($20 yearly) | Five-hour allowance, credits extend it |
| Enterprise, Edu | Contact sales | Flexible pricing scales with credits |
Weekly limits may also apply. OpenAI publishes message estimates per five hours, but they are estimates: the model, task size and tool use matter more than prompt length.
To check, run /status in a Codex CLI session, or open chatgpt.com/codex/settings/usage for limits, reset times and your credit balance.
At the limit, Codex can finish the current turn. After that, Plus and Pro can buy credits without upgrading, Business and Enterprise can buy workspace credits, and anyone can keep going with an API key at API prices.
ccusage: a usage tracker for your terminal
ccusage is a free, open-source (MIT) CLI that reads the session logs Claude Code and Codex keep on your machine and turns them into reports.
npx ccusage@latest # daily report
npx ccusage@latest monthly
npx ccusage@latest session # per conversation
npx ccusage@latest blocks # Claude Code 5-hour windows
npx ccusage@latest codex dailyIt runs on your machine, and your logs stay there. Two things to know. Its costs are estimates at API prices, so on Pro or Max they show what the work would have cost on the API, not your bill. And it only sees logs on this machine: claude.ai chats and other computers count toward your limit but aren’t in the report. So ccusage answers “how much did I use?” well. For “how much is left?”, /usage and the settings page are the source of truth.
What to do when you hit the limit
- Read the message. It says which limit you hit and when it resets.
- Model limit? Switch with
/model. Session or weekly limit? Switching won’t help. - Let Claude Code wait. Since v2.1.234, Claude Code can wait for the reset and continue the task by itself.
/rate-limit-optionsshows the choices. - Pay for more if it’s worth it.
/usage-creditsturns on usage credits (pay as you go past the limit)./upgrademoves you to a bigger plan. On Codex, buy credits. - Spend less next time.
/clearbetween unrelated tasks, Sonnet for routine work, lower effort with/effort, and follow the/usagerecommendations. - Switch agents. Your Claude and ChatGPT limits are separate. When one runs out, the other often still has room. See Codex vs Claude Code.
Know before you run out: Coral
The tools above tell you what you used. The harder question in the middle of a busy afternoon is: will I make it to the reset?
Coral is a Mac app that shows your coding agents in the Mac’s notch. Press U for what’s left of each plan: Claude Code, Codex, Copilot and Grok, read from each tool’s own account. When a limit will hit 0% before it resets at your pace so far, a line says when, like “Runs out Mon 23:00”. A limit that lasts says “Lasts to reset”, so a quiet Usage tab means you’re fine. Hover any limit to see its line anyway.
The iPhone app shows the same: the session and the week for each agent, when each resets, and a bar that turns amber below 25% and red below 10%. After time away, an amber segment shows what the week used while you were gone. Usage is part of Coral Free, on both. On Pro, Baton runs long and overnight tasks and stops before they eat your Claude week.
Usage drains fastest when several agents run at once. If that’s you, see Run several Claude Code agents at once.
Questions
How long is the Claude Code usage limit?
The session limit resets every five hours. The weekly limit resets once a week, at a fixed time assigned to your account on Max. Run /usage to see both reset times.
Do claude.ai and Claude Code share limits?
Yes. Claude, Claude Desktop and Claude Code all count against the same session and weekly limits on Pro and Max.
What’s the command to check usage in Claude Code?
/usage. It shows your plan bars, reset times and what used them. /cost and /stats open the same screen.
Does ccusage show my remaining limit?
No. It estimates use and API-price cost from local logs on one machine. Your remaining limit is in /usage or at claude.ai/settings/usage.
Does Codex have a weekly limit?
It can. OpenAI says weekly limits may apply on top of the five-hour allowance, and Pro plans currently have no five-hour limit. Check chatgpt.com/codex/settings/usage or /status.