OpenUsage
Track your AI coding subscriptions from the macOS menu bar.
OpenUsage shows how much of each AI plan you've used: session and weekly limits, credits, and
balances, all in one panel. The limits closest to running out sit right in the menu bar.
| Light | Dark |
|---|
| Priority | | |
| All Usage | | |
Installation
-
Homebrew:
brew install --cask openusage
-
Direct download: grab the latest universal DMG from the
releases page, open it, and drag
OpenUsage to your Applications folder.
The app updates itself with signed, notarized Sparkle updates. Requires macOS 15
(Sequoia) or later, on Apple Silicon or Intel.
Supported Providers
- Claude: session, weekly, model limits, extra usage (Enterprise spend), rate-limit resets
- ChatGPT: session, weekly, Spark, extra usage, credit limit, rate-limit resets, daily tokens
- Cursor: total usage, model usage, extra usage, requests, credits, daily spend and tokens
- Amp: agent usage, orb hours, Amp Free, credits
- Antigravity: Gemini and Claude quotas
- Copilot: credits, extra usage, chat, completions, organization billing, GitHub Enterprise on GHE.com
- Devin: daily and weekly quota, extra usage
- Grok: weekly limit, extra usage
- Muse Code: session, weekly
- Ollama: session, weekly, monthly, extra usage, daily usage and tokens
- OpenCode: session, weekly, monthly
- OpenRouter: balance, spend, key limit, budget
- Z.ai: session, weekly, web searches, rate-limit resets
- CLIProxyAPI hubs: every Claude, ChatGPT, Antigravity, Devin, and Grok login on
a CLIProxyAPI server
Add your accounts from the app's Accounts screen. OpenUsage never reads other tools' credentials,
files, or logs. See Accounts. Coming from 0.7.x? See
Upgrading from 0.7.x.
Features
- Menu bar. The limits closest to running out, at a glance.
- Priority. What's running low, what's used up, and which accounts need you.
- All Usage. Every account and limit in one place, with resets, your pace, and an optional mark
where an even pace would be.
- Your layout. Pin the metrics you care about, hide the rest with a right-click, and drag
everything into your order.
- Unify multiple accounts. Show your Claude or ChatGPT accounts as one card, with one bar per limit.
- Rate-limit resets. See and spend Claude and ChatGPT resets.
- Share. Copy a picture of your usage to the clipboard.
- Notifications. A heads-up before a limit runs out, or before a weekly limit resets with most of
it unused.
- Priority limits. Choose how much left turns a limit orange or red, see limits you're burning
through too fast while they still have plenty left, and see limits about to reset mostly unused.
- Global shortcut. Toggle the panel from anywhere.
- Command line. Your usage as JSON, for scripts and agents.
- Automatic updates. Signed, notarized updates, with an optional beta channel.
- Proxy support. Route provider requests through a SOCKS5, HTTP, or HTTPS proxy.
Documentation
How everything works, in docs/:
License
MIT
swift build # debug build
script/test_quiet.sh # run the tests without them taking the keyboard or the screen
./script/build_and_run.sh # build and launch the dev app from dist/
script/test_quiet.sh takes swift test arguments such as --filter. The dev app has its own
bundle identifier, so it runs beside an installed OpenUsage. See debugging.
To learn the code, read the architecture overview,
adding a provider, and AGENTS.md.
Create the GitHub Release for the tag as a draft with its notes, then push a v* tag on main to
build, sign, notarize, and publish it. A tag without a release that has notes fails before the build.
- A plain tag (
v0.8.0) ships to everyone.
- A pre-release tag (
v0.8.0-beta.1) ships to the beta channel.
The pipeline is release.yml, and the steps are in the
release-swift skill.
Release setup (one-time)
The workflow needs these repository secrets (Settings → Secrets and variables → Actions):
This is the beginning of the README. Read the rest at the original. Read the full README ↗
