Know exactly what
you spend on AI
Real-time Claude and OpenAI cost tracking from your menu bar. Limits and reset times for 30 providers, a forecast of when you'll hit them, and spend by model. No FavTray server stores your data.
Watch it in action
AI Usage: know when you'll hit your Claude limit. Limits, pace and the time you will run out for Claude, Codex and more, plus spend by model.
More videos →AI costs add up faster than you think
The average developer using Claude Code spends roughly $6 per day, with the top 10% exceeding $12 daily. Over a month, that's $120-250+ — and most developers have no idea until the bill arrives. OpenAI API usage can be equally surprising, especially when using GPT-4 or o1 for complex tasks.
FavTray puts your AI spending right in the menu bar where you can see it as you work. No more end-of-month billing surprises.
AI cost awareness without the overhead
Limits and Spend at a Glance
Keep your Claude or Codex session limit in the menu bar. One click shows each provider's limits and reset times, plus today's and the last 7 days' spend.
Claude & OpenAI Combined
Track both Anthropic Claude and OpenAI API costs in one place. Per-provider breakdowns show where your money goes.
Time-to-Limit Forecast
Each limit shows your pace and, at your current rate, when you'll hit it. Claude and Codex warn you at 80% and 95%, and FavTray notifies you when a limit runs out and when it's back.
No FavTray Server
No FavTray server stores your data. FavTray reads local log files and asks each provider's own API for your usage.
Spend by Model and Day
See which models your Claude Code and Codex spend went to, your input, output and cache token mix, and a day-by-day log. Claude Code tokens per project and per session, too.
Zero Configuration
FavTray reads ~/.claude/ logs automatically. No API keys to paste, no webhooks to set up, no accounts to create.
How FavTray compares to other trackers
| Tool | Price | Privacy | Setup | Providers |
|---|---|---|---|---|
| FavTray | Free | No FavTray server | Zero config | Claude + OpenAI |
| Helicone | Free tier / $20+/mo | Cloud-based | Proxy setup | Multi-provider |
| LangSmith | Free tier / $39+/mo | Cloud-based | SDK integration | LangChain only |
| OpenAI Dashboard | Free | Cloud-based | None | OpenAI only |
| Anthropic Console | Free | Cloud-based | None | Claude only |
Built for developers who use AI daily
Solo Developers
Track your personal AI spending without enterprise tools. See what each day costs and stay within your monthly budget.
Learn more →Freelancers
See what your AI work costs day by day, and which projects Claude Code spent its tokens on, before you quote or bill a client.
Learn more →Small Teams
Each developer runs FavTray on their own Mac and tracks their own limits and costs locally, without sharing data through a centralized platform.
Learn more →Cost-Conscious Engineers
See which models and days burn the most tokens, then tighten prompts and switch models where it counts.
Learn more →Frequently asked questions
Which AI providers does FavTray track?
FavTray tracks Claude (Anthropic) and OpenAI API costs in real time. It reads local usage logs and API data — no proxy setup or SDK integration required.
Is my API key safe?
Yes. API keys you add are kept in the macOS Keychain. No FavTray server stores your data: the app talks directly to each provider's own API, plus a license check and an update check.
How accurate is the cost tracking?
FavTray reads the same usage data that Anthropic and OpenAI use for billing. Costs are calculated using current per-model token pricing, updated with each app release.
Is the AI Usage Tracker free?
Yes, on every plan and with no daily limit: today's spend, 7- and 30-day totals, limits and the full usage history are all free. Pro, a one-time $4.99, unlocks other FavTray tools, not a better version of this one.
Do I need to set up anything?
No. FavTray automatically reads Claude usage from ~/.claude/ logs. For OpenAI, you provide your API key once. No webhooks, no proxy, no account creation.
Does my usage data leave my Mac?
Not to FavTray. Cost is worked out from the Claude Code and Codex log files already on your disk, and every chart is computed on your Mac. To read your limits, FavTray asks each provider's own API with the sign-in you already have. No FavTray server receives your usage, tokens or keys.
Stop guessing what AI costs you
Try FavTray free. Real-time AI cost tracking alongside eye rest and keep-alive — all in one menu bar app.