Tokenomics sits in your menu bar and watches every AI tool you pay for — so you know if you have room for one more run.
Claude
Pro
Coming soon — Midjourney, Suno, and Udio, once their public APIs ship.
SIGNAL
A ring at 60% means nothing on its own. Is that good or bad? It depends where you are in the window.
The pace dot shows where your fill would sit if you'd used the quota evenly since the window opened. If your ring is ahead of the dot, you'll run out of tokens. If it's behind, you have room.
That's the difference between data and signal. Tokenomics shows you both.
Understanding this signal helps you learn how prompts and models use tokens differently, so you can tune your session to fit your budget.
BREADTH
Most of us don't use just one AI tool. Tokenomics detects which ones you have installed and shows them all — usage rings in the menu bar, full detail in the popover, a dashboard on your desktop.
Drag to reorder. Hide the ones you're not using today. The menu bar shows the one you're closest to hitting.
Zero-friction authentication
Everyday AI tools store their credentials in different places. Tokenomics reads each one automatically: macOS Keychain for Claude, the gh CLI for Copilot, local configuration files for Cursor, OpenAI, and Gemini.
For providers that only give out API keys (Stability, Runway, ElevenLabs), paste once in Settings — that's the most friction you'll hit.
If Claude rotates your authentication token during a sleep/wake cycle — which it does — Tokenomics notices and re-reads Keychain silently. You won't see a logout screen.
Everyday AI
gh CLI · zero-friction
Creative AI
Widgets
Three sizes, same data. Mount them on your desktop and never click the menu bar again.
GitHub Copilot
Business
One provider, same rotation and pinning rules as the menu bar.
Up to three providers. Order matches the app — drag to rearrange.
Every provider you track, on one tile. Overkill for most. Glorious for the rest.
CRAFT
A fill for each provider ticks up as you burn through your budget. One look tells you where you stand.
Surfaces whichever provider is closest to its ceiling — the one that'll bite you first.
Skip the rotation. Lock the icon to the provider you actually care about.
Tells you whether to conserve — switch models, take a break — or whether you've got room to run.
Set a threshold per provider. Tokenomics pings you before the ceiling.
No data collected. No analytics. Nothing leaves your machine — ever.
Pricing
Enjoy Tokenomics on us — it's free forever.
That said, I'd love a coffee on you if you're up for it.
FREE FOREVER
Now and always, for anyone who installs today.
Tokenomics is free. If paid capabilities ever ship — a team or enterprise version is one possibility — the current app stays free forever for everyone who already has it.
Download for macOSBUY ME A COFFEE
Suggested Starbucks latte price.
Install
Sparkle handles silent background updates. No re-downloading from GitHub every time there's a new version.
macOS 14 Sonoma or later · Developer ID signed & notarized
Install Homebrew first, then run this command.
FAQ
Yes. Tokenomics is free to download and use. There's no subscription, no paid tier, and no feature gating. GitHub Sponsors link is in the app if you find it useful.
No manual setup required for most providers. Tokenomics reads credentials from where your AI tools already store them — macOS Keychain for Claude, the gh CLI for GitHub Copilot, and local configuration files for Cursor, OpenAI, and Gemini. Sign in to each tool once (as you normally would), and Tokenomics detects them automatically.
No. Your credentials are read from local sources into memory. They're never written to disk by Tokenomics, never logged, and never transmitted to any server other than the provider you're fetching usage from.
The only outbound calls Tokenomics makes are to Anthropic's API (for Claude usage) and to GitHub's appcast URL to check for app updates. OpenAI, Gemini, Copilot, and Cursor usage is read entirely from local files — no network calls.
You can read every line on GitHub. The license is source-available rather than MIT — it keeps commercial rights reserved so I have room to build paid capabilities later (a team or enterprise version is one possibility). The current app is free forever for anyone who installs it. This isn't a SaaS in disguise.
macOS only for now (14 Sonoma or later). The menu bar is a macOS-native concept and the auth paths (Keychain, gh CLI, local config files) are Mac-specific. A Windows/Linux version is on the roadmap. We love our Windows users too — we won't forget about you.
They're coming once those providers ship a public API with credit/usage endpoints. Tokenomics won't integrate against unofficial endpoints or scraping — that breaks with any platform update and it's against ToS. ElevenLabs, Runway, and Stability AI are next, because they have real APIs with documented quota endpoints.
You could. The problem is that you don't know you need to check until you're mid-task and something breaks. The pace dot is specifically useful here: it shows whether you're on track to hit a limit before you actually do, so you can make a call about whether to start a long agentic run.
A dashboard answers "how much have I used?" — Tokenomics answers "am I about to run out?"
Two things: breadth and signal quality. Breadth: eight providers in one place — five everyday AI providers (Claude, GitHub Copilot, Cursor, OpenAI, Gemini) plus three creative AI providers (Stability AI, Runway, ElevenLabs). Signal quality: the pace dot. Most usage meters show a percentage. Tokenomics shows whether that percentage is ahead of or behind where you'd expect given how much of the window has elapsed. That turns a number into an answer.
