Free limit calculator
Set how much of your 5-hour or weekly limit you've used and how long ago the window started. You'll see your pace and, if you're going too fast, when you'll run out. Works for Codex too.
Claude
updated just now
You've used 72% with only 55% of the window gone. At this rate you run out in about 1h 4m, 1h 11m before the reset.
In lpm, hover the usage meter in the sidebar to see this card while you work.
Your numbers stay on this page. In the lpm desktop app the same card fills itself in from Claude Code and Codex: a usage meter in the sidebar, a limits card for every Claude account, and the same meters in the lpm iPhone app.
The short answer
In Claude Code, run /usage to see your plan usage bars with reset times, the session's cost and a breakdown of what used your limits; /cost and /stats open the same screen. In Codex CLI, run /status for your plan's limits, usually 5-hour and weekly, as percent left, or /usage for token activity. To watch both in one place, by project and with a pace verdict, use lpm.
| Where | Claude Code | Codex |
|---|---|---|
| In a session | /usagePlan bars, session cost and a breakdown. /cost and /stats are aliases. | /statusLimits as percent left, with reset times. /usage shows token activity. |
| Always on screen | rate_limitsIn your statusline, on Pro and Max | /statuslinePin the rate-limit items to the footer |
| In the browser | claude.ai Settings > Usage | chatgpt.com/ |
| With an API key | Session tokens and cost, no plan bars | Billed at API rates, no plan windows |
| In lpm | Usage card and sidebar meter, after one click | Usage card and sidebar meter, no setup |
To keep 5-hour and weekly usage under every Claude Code prompt without editing config files, use lpm's statusline editor.
Both agents count usage against a short window and a weekly one. Here's what each plan does, with links to the rules.
Checked . Allowances change; the linked pages have the current rules.
It divides the share of your limit you've used by the share of the window that has passed. Above 1.15 you're going too fast, and it projects your current rate forward to the moment you'd reach 100%. If that comes before the reset, it shows how long you have left and how early you'd hit the limit. Below 0.85 your usage will last, and in the first 5% of a window there's no verdict yet.
No. It only uses the numbers you set, and they stay on this page. To fill them in, run /usage in Claude Code or /status in Codex. The lpm desktop app reads the same percentages from Claude Code and Codex on your computer and keeps the pace up to date as you work.
lpm divides the share of a window you've used by the share of time that has passed. Above 1.15 reads “too fast”, below 0.85 reads “will last”, and anything between is “on pace”; in the first 5% of a window there's no pace verdict, only “limit reached” at 100%. When it reads too fast and the current rate would use up the window before it resets, the meter adds an estimate such as “runs out in ~1h 4m, before reset”.
On a Claude subscription, usage counts against a session window that resets every five hours and a weekly limit across all models. Both are shared with claude.ai, Claude Desktop and the IDE extensions, and Anthropic doesn't publish a token number for either. Max 5x and Max 20x give five and 20 times Pro's per-session allowance. An API-key login has no plan windows: you pay per token, and lpm's Claude card says there are no limits to show.
At a fixed day and time each week that Anthropic assigns to your account, not seven days after you start. /usage and claude.ai Settings > Usage show the next reset. In lpm, the Weekly meter shows it as a countdown and a date, such as “resets in 3d 19h · Sep 27, 4:30 PM”, and the sidebar meter shows the weekly window by default.
In Codex CLI, run /status. It lists the limits your plan reports, usually a 5-hour and a weekly window, as percent left with reset times. /statusline pins limit items to the footer, and /usage shows daily, weekly or cumulative token activity when you're signed in with ChatGPT. On the web, open chatgpt.com/codex/settings/usage. lpm reads Codex limits with no setup and shows them as percent used, beside Claude's.
Wait for the reset, use a free reset if you have one, or pay to keep going. Claude Code 2.1.234 and later can wait in the open session and continue the interrupted task after the reset. Anthropic occasionally gives eligible plans a free limit reset, used from Settings > Usage on claude.ai or in Claude Desktop, and Pro and Max can turn on usage credits or upgrade. When Codex hits a limit, the current turn finishes; you can redeem an earned reset from /usage if you have one, Plus and Pro can buy credits, or you can switch to an API key. In lpm, press ⌥↵ (Alt+Enter on Windows and Linux) and choose Limit resets: your next prompt goes out 30 seconds after the window resets, once the agent is idle, in a local Claude Code or Codex terminal.
Tokens by project and session, estimated cost, and live 5-hour and weekly meters with this same pace verdict.
Give each project its own Claude account, and move new sessions to one with room when an account runs low.
Keep 5-hour and weekly usage under every Claude Code prompt, or pick Codex's limit fields.
lpm is free and open source. It reads your 5-hour and weekly limits from Claude Code and Codex and keeps the pace in view while your agents work.