Home / Guides / Usage monitors compared

Claude Code usage monitors for Mac, compared (2026)

Updated 1 Oct 2026 by DevSparks

How do Claude Code limits work?

Claude Code on a Pro, Max, Team or seat-based Enterprise plan is metered by two limits at once: a five-hour session limit and weekly limits. The Claude Help Center describes both on the Settings > Usage page, which shows how much of the five-hour session you have used, the time left in it, and when your weekly limit resets (with a separate bar for Fable if your plan includes it).

The allowance is shared. Anthropic says your usage of claude.ai, Claude Code and Claude Desktop all counts towards the same limit, so a long chat in the browser eats into the budget your terminal sessions need. How fast you use it depends on message length, files, tools, the model and the effort level.

Claude Code also hands these figures to your status line script. Its documentation lists rate_limits.five_hour and rate_limits.seven_day, each with a used percentage and a reset time. The field appears only for claude.ai Pro and Max subscribers, and only after the first reply in a session.

How do Codex limits work?

Codex, when you sign in with ChatGPT, uses a 5-hour window and a weekly window too. OpenAI's help article on using Codex with a ChatGPT plan refers to both "5-hour and weekly Codex usage windows", and the Codex pricing page says local messages and cloud chats share your plan's allowance, with weekly limits that "may also apply".

The 5-hour window has not been constant. On 24 August 2026, OpenAI's Codex lead said the 5-hour limit would return for Plus accounts the next day, and that it would stay off "for the upcoming months" on Pro $100 and Pro $200. Check /status rather than assuming which windows your plan has today.

What can you check without installing anything?

• Claude Code /usage: shows session cost, plan usage limits and activity stats. /cost and /stats are aliases.
• claude.ai Settings > Usage: progress bars for the five-hour session and weekly limits, plus usage credits.
• Codex /status: OpenAI's pricing page says it shows your remaining limits during an active Codex CLI session.
• The Codex usage dashboard at chatgpt.com/codex/settings/usage: current limits, credit balance and reset times.

All four are official, and all four answer only when you ask. None of them runs in the background or warns you before a long task starts. The full walkthrough is in how to check your Claude Code and Codex usage limits.

Which usage monitors are there?

Fourteen tools and the built-in checks, compared on what each one's own site, README or store listing says. Where a listing does not say, the table says "not stated". Prices are as listed on 1 Oct 2026.

Two tools share the name Claude Usage Tracker: the browser extension by lugia19 and the Mac app by hamed-elfayome. The table keeps them apart. ClaudeOps, which some lists still mention, is left out: its site returned "not found" on 1 Oct 2026.

Which one should you pick?

It depends on what you want to see and what you are willing to hand over. Going by what each tool says about itself:

• A free number in the menu bar: Claude God and ClaudeUsageBar are free and open source; CodexBar and the Claude Usage Tracker Mac app are MIT licensed.
• A view inside the terminal: Claude Code Usage Monitor for a live meter, ccusage for token and cost reports.
• Many AI tools in one place: SessionWatcher (20 tools), Code Meter (14 providers) and CodexBar.
• Limits on your phone or watch: AI Limits Tracker and Usage for Claude.
• No credential handed to the app: tools that read only local files or Claude Code's status line data, which are Claude Code Usage Monitor, ccusage and Vibepanions.
• Several sessions at once, and you need to know which one is waiting: Vibepanions shows every session and takes you to its terminal. The Claude Usage Tracker Mac app has a beta Dynamic Island that pulses when Claude Code needs input. No other tool here says it does this.

Why limits are only half the problem

The limit tells you how much you have left; it does not tell you which session is burning it or which one stopped to ask you something. With four or five Claude Code and Codex sessions across terminal tabs, the usual failure is not running out. It is a session that sat for twenty minutes waiting for a yes, while the 5-hour window kept moving.

Most monitors in the table are meters. They answer "how much", not "where". If you run several sessions at once, look for a tool that also shows session state and can take you to the right window.

How Vibepanions handles it

Vibepanions is a macOS app that shows how much of your Claude Code and Codex usage limit is left, when it resets, and which terminal session is waiting for you. It floats above every window as a small widget with a ring for the 5-hour limit (Claude or Codex, whichever the office shows), and opens into an office view with one desk per session.

• Claude limits: Claude Code reports your plan's 5-hour and weekly usage to its status line. Vibepanions installs a small pass-through in front of your own status line and shows those real figures. Until Claude Code has reported, it estimates from your own past windows and labels the numbers "estimated".
• Codex limits: read from what Codex writes into its session files, so they are reported, not estimated.
• Sessions: desks turn green while working and amber when a session waits for you. The widget waves, shows an amber badge, and can send a notification.
• Focus: click a desk to jump to the exact terminal in iTerm2, Terminal, Cursor, VS Code, Warp or Ghostty.

There are also fifteen unlockable companions, the first (Miso) at 200,000 characters. They are there to keep the window pleasant to leave open; the limits are the point.

What should a monitor be allowed to read?

As little as it needs. The tools above get their numbers in four ways: local log files, Claude Code's status line data, the login Claude Code keeps in your Keychain (used to ask Anthropic's usage API directly, as Usagebar, Limits Monitor and Claude God describe), or your signed-in claude.ai account (OAuth, browser cookies or a pasted session key). The first two never leave your Mac. The last two mean the app holds or uses a credential that can act as you, so check what it stores and where it sends it.

Vibepanions reads only local files in ~/.claude, ~/.codex and ~/.vibepanions, and never asks for a Claude or OpenAI password. Optional GitHub sign-in, to keep your score online, shares only your login, avatar, character totals, a character id and a coarse state, never prompts or code. The details are on the privacy page.

Five habits that stretch your limit

Anthropic's own guide to Claude Code usage lists five habits behind most "I burned through my limit by lunchtime" reports:

1. Clear between tasks. Every earlier message is resent on every turn, so run /clear when you switch problems.
2. Match the model to the job. Sonnet for most coding, Opus when you are stuck, Haiku for quick mechanical work.
3. Point at files instead of pasting them. A bare path lets Claude read only the part it needs.
4. Keep CLAUDE.md lean. It is sent with every turn; Anthropic suggests under roughly 200 lines.
5. Ask for a plan before big changes. A plan costs a few hundred tokens; a wrong 400-line diff costs thousands.

What happens when you hit the limit?

On Claude, you wait for the reset, upgrade, or buy usage credits. Usage credits are available on Pro, Max 5x and Max 20x, are billed at standard API rates, cover both claude.ai and Claude Code, and have a daily redemption limit of $2,000. The five-hour limit keeps resetting as usual.

On Codex, OpenAI says a turn that is already running can finish, subject to fair-use limits. After that, the usage page shows your options: add credits, apply an available reset, upgrade, or wait for the reset.

FAQ

Sources

Fetched 23 Sep 2026. The eight tools added on 1 Oct 2026 were fetched that day.

• Claude Help Center, Usage limit best practices
• Claude Help Center, How do usage and length limits work?
• Claude Help Center, Manage usage credits for paid Claude plans
• Claude Help Center, Models, usage, and limits in Claude Code
• Claude Code docs, Status line (rate_limits fields)
• Claude Code docs, Commands (/usage)
• OpenAI Help Center, Using Codex with your ChatGPT plan
• OpenAI, Codex pricing
• 9to5Mac, OpenAI restores 5-hour Codex and Work limits for ChatGPT Plus users (24 Aug 2026)
• Claude Code Usage Monitor README
• ccusage README
• Claude Usage Tracker extension README
• CodexBar README
• Claude Usage Tracker (Mac app) README
• Usagebar
• SessionWatcher
• Code Meter, Mac App Store listing
• ClaudeUsageBar
• Limits Monitor for Claude Code
• AI Limits Tracker
• Claude God
• Usage for Claude, App Store listing

Next: How to check your Claude Code and Codex usage limits.