Claude Code model router

Tokens 4 Breakfast

Track AI usage, spend, subscriptions and rate limits across 11 providers (Claude Code, Claude, Codex, Cursor, Copilot, OpenAI, Anthropic API, Grok, DeepSeek, Mistral, OpenRouter) in your Mac menu bar. Local-first, no telemetry.

Install

Install
brew install --cask onekapisch/tap/tokens-4-breakfast

Found in the project README. Check it for setup steps after installing.

From source
git clone https://github.com/onekapisch/Tokens-4-Breakfast.git

From the README

AI spend, tokens and rate limits across 11 providers, in one private Mac menu bar app.
Claude Code · Claude · Codex · Cursor · GitHub Copilot · OpenAI · Anthropic API · Grok · DeepSeek · Mistral · OpenRouter

Tokens 4 Breakfast is a commercial, closed-source macOS app. This repository hosts the public docs and the issue tracker, not the source code. Bug reports and feature requests are welcome in Issues.

Why

If you build with AI, your spend and your limits are spread across a handful of tools, and you usually find out about either one too late:

- The bill. Claude, ChatGPT, Cursor, Copilot and a few API keys each bill separately, and nothing adds them up until the invoices arrive.
- The limit. You hit a Claude or Codex window halfway through an agent run, with no warning.

Tokens 4 Breakfast keeps both numbers in your menu bar while you work: what your usage is worth, what your subscriptions cost, and how much of each limit you have left.

A look inside

      Limits. Every window with its reset time, a pace marker and a forecast.

      Insights. Tokens and their API-equivalent value at a glance.

      Plans. Whether your subscriptions pay for themselves.

Screenshots are rendered from the app with sample data, not a real account.

What you get

- Live numbers in the menu bar. Show a provider's cost, all-provider spend, a subscription total, a Claude 5-hour or weekly cap, or Codex and Grok usage, with up to two extra compact metrics beside it.
- Know your pace. Every 5-hour and weekly meter marks where an even pace would be, with one word beside it: *burning fast*, *on pace* or *room to spare*. 90% used with an hour to go reads very differently from 90% with two days left.
- Limits before you hit them. Claude 5-hour, weekly and per-model limits, Codex limits and…
Read the full README on GitHub

Details

Last updated
7 November 2026
Topics
aiai-spendccusageccusage-alternativeclaude-codecost-trackingcursordeveloper-toolsgithub-copilotllmlocal-firstmacos

FAQ

What is Tokens 4 Breakfast?

Track AI usage, spend, subscriptions and rate limits across 11 providers (Claude Code, Claude, Codex, Cursor, Copilot, OpenAI, Anthropic API, Grok, DeepSeek, Mistral, OpenRouter) in your Mac menu bar. Local-first, no telemetry. It is a model router for Claude Code, from onekapisch/Tokens-4-Breakfast on GitHub.

How do I install Tokens 4 Breakfast?

Run brew install --cask onekapisch/tap/tokens-4-breakfast, then follow the setup in the README.

Is Tokens 4 Breakfast popular?

onekapisch/Tokens-4-Breakfast has 26 GitHub stars and 0 forks. It was last updated on 7 November 2026.

More model routers & proxies

Browse the most popular Claude Code mods, all model routers & proxies, or the skills and MCP server directories.