At-a-glance tray icon
Your session usage on a color-coded ring right next to the clock, with the reset countdown in the tooltip.
Claude usage tracker for Windows 10 & 11
UsagePeek AI lives in your Windows tray and shows your Claude usage in real time: current session, weekly limit and per-model caps, so you never hit a wall mid-conversation.
Native tray app · Local-only · Zero telemetry · 10 languages
From a glance at the clock to a full breakdown, pick the level of detail you want.
A color-coded ring beside the clock. Hover for the reset countdown, no clicks needed.
Session, weekly and per-model usage with exact reset clocks, plus your local Claude Code spend. Four layouts, six themes.
An always-on-top mini view you can drag and resize. Three designs: bars, ring badge or half-circle gauge, and you choose which per-model limits ride along: Fable, Opus, Sonnet, Design.
What's new in v1.16
Pick the floating widget's main metric: session, weekly, or a per-model cap like Fable. It takes the top bar, the outer ring or the big number on the gauge, in all three designs. Everything else stays right behind it, and a per-model cap only shows up when your plan reports it.
A focused utility that does one job extremely well.
Your session usage on a color-coded ring right next to the clock, with the reset countdown in the tooltip.
Session, weekly and per-model usage with countdowns and exact reset times. Four layouts, six themes.
An always-on-top mini view you can drag and resize. Three designs: bars, ring badge or half-circle gauge. Pick which per-model limits it shows and which one leads, or keep it down to session and weekly.
When your limit resets, your queued projects relaunch automatically: in a terminal with claude --resume or straight into Claude Desktop. Wake up to work already resumed.
Log in through a built-in window. Your session token lives in Windows Credential Manager, never in a file.
Local-only, zero telemetry. It only talks to claude.ai and can't read your chats, projects or files. Available in 10 languages.
Estimates the cost of your Claude Code sessions from the logs on your PC, per day and per model family, with zero extra requests. Charted in its own statistics window.
The same app for the Mac menu bar, with feature parity: popup, widget and local Claude Code stats. In private beta now.
macOS · coming soon
UsagePeek AI for macOS is in private beta, mirroring the Windows app: same popup, same floating widget, same auto-resume and the same local Claude Code statistics.
Get notified when it shipsUsagePeek AI has no server of its own. Nothing about you (not your token, not your usage, not your settings) ever leaves your machine for anywhere but claude.ai's own usage endpoints.
Read the privacy policyYou sign in once through the app's built-in login window. The resulting session token is stored in Windows Credential Manager, never in a file, and is used only to call Claude's own usage endpoints. The app talks to claude.ai and nothing else.
No. UsagePeek AI only reads usage numbers: how much of your session, weekly and per-model limits you have used. It cannot access your conversations, projects or files, and it contains zero telemetry.
Yes. The limits UsagePeek tracks are the same ones Claude Code draws from. It also estimates your Claude Code spend locally, charts it per day and per model family in a statistics window, and has auto-resume: when your limit window resets, it relaunches the projects you queued, in order, each in a terminal or in the Claude Desktop app.
From the session logs Claude Code already keeps on your PC. UsagePeek reads the token counts per message, applies a published price table, and aggregates them per day and per model family. It makes zero extra network requests for this: the estimate never leaves your machine.
Because your plan is not reporting one. UsagePeek shows exactly the buckets claude.ai itself reports for your account: the 5-hour session, the weekly limit across all models, and a per-model weekly cap only for the models your plan caps separately. Everything without its own cap, Opus included on most plans, counts towards Weekly · all models. Nothing is missing or hidden: if a cap starts being reported for your account, its row appears on its own.
Session and weekly always, plus any per-model cap you tick under Settings → Floating widget: Fable, Opus, Sonnet and Claude Design. The list is separate from the popup's, so you can keep the widget down to two numbers while the popup shows everything. As everywhere in the app, a model only shows up when your plan reports that cap. The same section lets you pick the main metric, for example Fable weekly, which then takes the top bar, the outer ring or the big number.
Coming soon. UsagePeek AI for macOS is in private beta, living in the menu bar with feature parity: popup, floating widget, auto-resume and the local Claude Code statistics. Write to [email protected] if you want to be notified when it ships.
Not anymore. Since v1.7, every project in the queue chooses where it comes back: a Windows Terminal running claude --resume, or the Claude Desktop app, which opens the very same session. Mix both in one queue if you like.
Windows 10 and Windows 11. UsagePeek AI is a native tray app, distributed through the Microsoft Store.
No. UsagePeek AI is an independent, third-party utility. It is not affiliated with, endorsed by, or sponsored by Anthropic. "Claude" and "Anthropic" are trademarks of Anthropic, PBC.
UsagePeek AI is distributed through the Microsoft Store. See the Store listing for current pricing and trial options. Purchases and licensing are handled entirely by Microsoft.
Yes. Claude Usage Monitor is the browser-extension sibling of UsagePeek AI, built by the same developer, for Chrome and Firefox. If you spend more time in the browser than in the tray, it's a great fit.
Put your Claude limits where you can see them, right next to the clock.
Get it from theMicrosoft StoreFor Windows 10 & 11