Coffee
← Blog AI Cost Management

ccusage Alternatives for Mac in 2027

By FavTray ·

Written by the team behind FavTray, a Mac developer dashboard for the menu bar. We check every app we cover against its own site, docs or a hands-on test, and each post says which.

Short answer: ccusage is a free, open-source CLI that reads your coding agents’ local logs and estimates tokens and cost. It can’t see plan limits and only runs when called. For a ccusage alternative on a Mac, a menu bar app such as FavTray AI Usage (free), Claude Usage Tracker (free), SessionWatcher or Usagebar shows limits, resets and spend all day.

Checked 14 October 2026 against the ccusage GitHub repo, ccusage.com docs, the npm registry and each app’s own site.

TL;DR

  • ccusage (v20.0.33, MIT) reports tokens and estimated cost from local logs for Claude Code, Codex and 15+ other agents.
  • It doesn’t know your plan limits. blocks approximates the 5-hour window; it can’t show the weekly limit or a real percentage.
  • No live view since blocks --live was removed in v18. The statusline command (beta) is the closest thing.
  • Built-in first: /usage in Claude Code and /status in Codex show the real limits for free.
  • Menu bar alternatives add limits, resets, alerts and history without opening a terminal. We make one of them (FavTray), so weigh our take accordingly.

What is ccusage?

ccusage describes itself as a tool to “analyze coding (agent) CLI token usage and costs from local data”. It’s MIT-licensed, has about 19,000 GitHub stars, and npm counted roughly 365,000 downloads in the week to 12 October 2026. The latest release on npm when we checked was 20.0.33, published 14 October 2026.

You don’t install anything permanent:

npx ccusage@latest            # daily report, all agents
npx ccusage@latest monthly    # by month
npx ccusage@latest session    # by conversation
npx ccusage@latest blocks     # Claude Code 5-hour blocks
npx ccusage@latest codex daily  # one agent only

bunx ccusage and pnpm dlx ccusage work too. Useful flags include --since and --until for a date range, --breakdown for per-model costs, --project (with ccusage claude) to filter one Claude Code project, --json for scripts and --offline to skip fetching prices.

It now covers far more than Claude Code: Codex, OpenCode, Amp, Droid, Gemini CLI, GitHub Copilot CLI, Goose, Kimi, Qwen and others, each with its own subcommand.

It’s maintained by ryoppippi and released often, so @latest matters: an old cached copy may not have prices for a model that launched last week, so its cost figures for that model can be off. If you run ccusage from a script, pin a version and update it on purpose, or pass --offline so the result doesn’t depend on a network fetch. For one-off checks, npx ccusage@latest is the simplest way to stay current.

What does ccusage read?

Only files already on your Mac. According to the ccusage guide:

AgentWhere ccusage looks
Claude Code~/.config/claude/projects/ and ~/.claude/
Codex$CODEX_HOME, or ~/.codex if it isn’t set
OthersEach agent’s own local session folder; custom paths via flags

It adds up the tokens in each logged request and multiplies by model prices from the LiteLLM pricing table. With --offline it uses a cached copy of those prices. ccusage says your usage data never leaves your computer.

What are ccusage’s limitations?

None of these are bugs. They follow from reading logs instead of asking the provider.

LimitationWhy it matters
No plan limitsLogs don’t contain Anthropic’s or OpenAI’s limits, so ccusage can’t say “you’ve used 72% of this session” or anything about the weekly limit
Blocks are approximateA block starts at your first message and lasts exactly five hours, calculated in UTC. Usage on claude.ai or another Mac doesn’t appear, so it won’t match the official window exactly
Cost is an API-price estimateOn Pro or Max you pay a flat fee, so the dollar figure shows value used, not money spent. Prices come from LiteLLM, not your invoice
No live monitorblocks --live was removed in v18.0.0. You run a command each time, or use the beta statusline
No alertsNothing tells you when you’re close to a limit
One machineIt only sees logs on the Mac it runs on

We cover the “why do the numbers differ” question in more depth in how to see Claude Code usage limits on Mac.

Do you need an alternative at all?

Maybe not. Two built-in commands and one ccusage feature cover a lot:

  • /usage in Claude Code shows your 5-hour and weekly plan usage and when each resets. It’s the official number.
  • /status in Codex CLI shows your remaining Codex limits during a session. Our guide to Codex usage limits has the details.
  • ccusage statusline puts session cost, today’s cost, the current block’s cost and time left, burn rate and context use into Claude Code’s status line. Add this to ~/.claude/settings.json:
{
  "statusLine": {
    "type": "command",
    "command": "bun x ccusage statusline",
    "padding": 0
  }
}

That’s enough if you only care while a session is open. A menu bar app earns its place when you want the numbers when no terminal is open, want a warning before you’re cut off, or juggle Claude and Codex.

What are the best ccusage alternatives for Mac?

AppBest forPriceFree tier?
FavTray AI UsageClaude, Codex and other providers’ limits plus spend by model, in one menu bar appFree in full; Pro $4.99 once for other toolsYes, full
Claude Usage TrackerA free, open-source Claude and Codex limits appFree (MIT)Yes
SessionWatcherMany providers, widgets, iPhone and Watch companion$6.99 to $59 one-time, or $24/yearNo
UsagebarA small Claude Code limits bar$9 one-timeYes, limited
Claude Code Usage MonitorA live terminal dashboard with forecastsFree (MIT)Yes

FavTray AI Usage

FavTray is a menu bar toolkit, and its AI Usage tracker is free on every plan with full history and no daily limit.

Good:

  • Your Claude 5-hour session, weekly limit and any model-specific limit, plus Codex session and weekly limits, with reset times. The same screen covers about 30 providers, including Cursor, Gemini and Copilot.
  • A pace marker on each bar and a forecast of when you’ll hit the limit at your current rate.
  • Notifications at 80%, 95% and when a limit runs out, and again when it’s back.
  • Spend today, over 7 days and 30 days by model, worked out from the same local Claude Code and Codex logs ccusage reads. Claude Code tokens per project and per session too.
  • It reuses the Claude Code login already on your Mac, read-only, and never refreshes it, so it can’t sign you out. No FavTray server sees your usage.
  • A percentage or reset countdown as text in the menu bar, and the same bars in the notch with FavShelf.

Not great:

  • No CLI or JSON export of the usage data. If you pipe ccusage into scripts, keep ccusage for that.
  • Fewer agents than ccusage for cost from logs: cost is Claude Code and Codex.
  • Saving and switching between several Claude Code or Codex accounts is a Pro feature.

Price: free. Pro is $4.99 once and covers other FavTray tools, not a better AI Usage. macOS 14 Sonoma or later.

Claude Usage Tracker

A native Swift menu bar app focused on Claude, with Codex added in v3.3.0 (August 2026).

Good:

  • Free and open source (MIT), with about 3,700 GitHub stars and recent releases.
  • Shows the 5-hour session, weekly and per-model limits; Codex windows map onto the same rows.
  • Multiple Claude accounts and a terminal status line.

Not great:

  • It tracks Claude API console costs, but has no cost-from-logs view like ccusage’s.
  • Claude and Codex only.

Price: free, with an optional Buy Me a Coffee. macOS 14 or later.

SessionWatcher

A paid tracker that covers 20 tools across Mac and Windows.

Good:

  • 5-hour and weekly limits with reset times, usage-rate charts and token and cost breakdowns.
  • Notifications at 70% and 90% by default, background refresh about every 90 seconds.
  • Pro adds macOS widgets, 90 days of history, multiple Claude and Codex accounts and up to 5 devices. There’s a beta iPhone and Apple Watch app.

Not great:

  • Paid only after a 7-day free trial, and the cheapest tier covers one tool.
  • Prices rise on 1 November 2026, according to its site.

Price: Solo $6.99 (one tool), Bundle $14.99 (Claude, Codex and Cursor), Pro $59 lifetime or $24 a year. 14-day money-back guarantee. macOS 14 or later.

Usagebar

A single-purpose Claude Code limits bar.

Good:

  • Reads Claude Code’s login from your Keychain and calls Anthropic’s usage endpoint directly. No separate account.
  • 5-hour and weekly limits, reset timers and notifications.

Not great:

  • Claude Code only, as far as its site says.
  • No spend or token history.

Price: $9 one-time with lifetime updates. Its site mentions a free tier without listing what it includes.

Claude Code Usage Monitor

The closest thing to “ccusage, but live”, in the terminal.

Good:

  • A live, colour terminal dashboard with burn rate and forecasts.
  • Can capture Claude Code’s official rate limits through a status line hook, and falls back to clearly labelled estimates.
  • Free and MIT. Install with uv tool install claude-monitor.

Not great:

  • Still a terminal window you have to keep open.
  • Its per-plan token figures for Pro, Max 5x and Max 20x are its own estimates, not numbers Anthropic publishes.

Price: free.

How do menu bar apps get your real limits?

This is the part ccusage can’t do, so it’s worth knowing how the alternatives manage it, and what to check before you hand one your login.

Logs give cost, the provider gives limits. Every app on this list that shows a real percentage of your 5-hour or weekly limit gets it from the provider, not from your logs. For Claude, that means reading the Claude Code login already stored on your Mac (macOS keeps it in a Keychain item) and asking Anthropic for your plan usage, the same figure /usage shows. For Codex, it’s the ChatGPT sign-in Codex saved, and OpenAI’s answer. Cost and token history still come from local logs, the same files ccusage reads.

What to check before you install one:

QuestionWhy it matters
Does it only read your existing login?An app that refreshes or rewrites Claude Code’s or Codex’s credentials can sign the CLI out or trigger repeated logins. Read-only is safer
Who does it talk to?Ideally only the provider’s own API. A third-party server in the middle sees your usage, and possibly your token
How often does it refresh?Every few minutes is plenty. Very frequent polling adds requests to your account and battery use for no gain
Does it need an API key?For plan limits it shouldn’t. Keys are for API-billed spend, and belong in the Keychain
Is it maintained?Providers change their usage endpoints. An app without a release in months may quietly stop showing numbers

macOS will ask once. The first time an app reads Claude Code’s saved login, macOS shows a Keychain prompt for the “Claude Code-credentials” item. Choose Always Allow only for an app you trust; Allow asks again next time.

Numbers won’t match ccusage exactly, and that’s fine. ccusage’s block starts at your first logged message and only sees this Mac. The provider’s window follows its own clock and counts claude.ai, the desktop app and other machines too. When they disagree, the provider’s percentage is the one that decides whether you get cut off.

Which one should you pick?

You want…Use
Cost per project or per day, in scripts or CSVccusage
The official percentage of your limit, now and then/usage (Claude Code) or /status (Codex)
Cost in Claude Code’s status line while you workccusage statusline
Limits and spend all day without a terminal, freeFavTray AI Usage or Claude Usage Tracker
Many providers, widgets and a phone app, and you’ll paySessionWatcher
A live terminal dashboardClaude Code Usage Monitor

Can you use ccusage and a menu bar app together?

Yes, and plenty of people do. They answer different questions. The menu bar app answers “how close am I to being cut off, and when does it reset?” ccusage answers “what did last month’s work on this repo cost at API prices?” Both read the same local logs, and neither changes them.

If the numbers worry you, our guide to reducing Claude Code token usage is the next step, and the best AI usage tracking tools covers API-spend dashboards beyond coding agents.

FavTray’s AI Usage tracker is free with no trial needed; the 7-day Pro trial (no card) covers the rest of the app. See pricing for Pro.

Sources (checked 14 October 2026): ccusage on GitHub, ccusage guide, blocks reports, status line guide, the npm registry, and each app’s page linked above.

Try FavTray Free

One free Mac menu bar app: notch island, ports, keep-awake, clipboard, eye rest, and Claude and Codex limits.

Download Free

Frequently Asked Questions

What is ccusage?

ccusage is a free, open-source (MIT) command-line tool that reads the log files coding agents write on your machine and reports tokens and estimated cost by day, week, month, session or 5-hour block. It started with Claude Code and now also covers Codex, Gemini CLI, GitHub Copilot CLI, OpenCode, Amp and more. Run it with npx ccusage@latest.

Is there a GUI for ccusage?

ccusage itself is terminal-only, with a beta status line for Claude Code. For a graphical view, use a menu bar app that reads the same local logs and also asks the provider for your plan limits. FavTray's AI Usage tracker is free with full history, Claude Usage Tracker is free and open source, and SessionWatcher and Usagebar are paid.

Does ccusage show my Claude Pro or Max usage limit?

No. ccusage works from local logs, which don't contain Anthropic's plan limits, so it can't tell you what percentage of your 5-hour or weekly limit you've used. Its blocks report groups usage into 5-hour windows that start at your first message, which approximates the session window. Use /usage in Claude Code, claude.ai or a limits app for the real percentage.

Is ccusage cost accurate on a Claude subscription?

It's an estimate of what the same tokens would cost at API prices, using model pricing from LiteLLM. On a Pro or Max subscription you pay a flat monthly fee, so the figure shows value used rather than money spent. On API billing it's a close guide, but your Anthropic or OpenAI invoice is the final word.

Does ccusage send my data anywhere?

ccusage says your usage data never leaves your computer. It reads logs locally and calculates cost on your machine. It may fetch model pricing data, and the --offline flag uses cached pricing instead, for air-gapped machines. Menu bar alternatives that show plan limits do make one request: to the provider's own API, with the login you already have.