ccusage Alternative: A Visual, Always-On Desktop GUI

The short version

ccusage is a fast, open-source command-line tool that turns your local Claude Code (and Codex) logs into daily, monthly, per-session, and 5-hour billing-block cost reports right in your terminal. Token Forest is an independent desktop app that reads those same logs and turns them into a cozy pixel pet plus a three-view dashboard. Neither uploads anything; both estimate cost locally.

Key takeaway: Token Forest is a friendly visual alternative to ccusage — not a front-end for it. It shares no code with ccusage and works entirely on its own. If you love the terminal, keep using ccusage; if you want an always-on, at-a-glance view, add Token Forest.

What ccusage does well

ccusage (npm package "ccusage", by ryoppippi, MIT-licensed) is a genuinely great tool, and we recommend it for terminal-first workflows.

• ✦Instant reports in any terminal — run it with npx, no install required.
• ✦Daily, weekly, monthly, per-session, and 5-hour billing-block breakdowns of tokens and estimated cost.
• ✦It focuses on terminal reports, plus an experimental (Beta) status-line display for Claude Code.
• ✦Scriptable and CI-friendly — JSON output, date filters, per-model breakdowns.
• ✦It reads many coding agents (15+ including Claude Code — Codex, Gemini CLI, Copilot CLI, and more), not just one.

What Token Forest adds

Token Forest is a cozy pixel-art desktop pet for Windows 10/11 and macOS. It's free (public beta), local-first, and works fully offline.

• ✦An always-on companion: your token usage grows a living pixel tree through 8 stages while you work — no command to run.
• ✦A full offline dashboard with three reconciling views — Growth (per-tree stage, days planted, “this tree ≈ $X”), Usage (daily/weekly/monthly charts, a per-model breakdown for Claude Code and Codex, a 26-week heatmap, burn rate, and per-project totals), and Chats (one line per conversation with its four token classes and estimated cost).
• ✦Gentle, gamified motivation instead of raw numbers — watch effort turn into something that grows.
• ✦An optional global leaderboard that is off by default; you opt in only if you want to compare.

ccusage vs Token Forest at a glance

Both read the same logs — and never your code

Both tools read only the local usage logs on your machine — for Claude Code at ~/.claude and for Codex at ~/.codex. Token Forest uses only token counts plus a little structural metadata (model name, timestamp, session title, project-folder name, git branch, session id, and message-type structure). It never reads your source code, your prompts, or the content of your conversations, sends no telemetry, and runs with the network fully disabled.

Why the token numbers look huge

Both tools track four token classes: input, output, cache-read, and cache-write. Cache-read tokens dominate the raw counts but cost very little each — in Token Forest's own six-week sample, cache reads were roughly 57% of real cost while input and output together were only about 24%. That's why a raw token total can look enormous while the real cost stays modest.

One honest caveat for both tools: the dollar figures are estimates computed locally from a price table, and they can differ from your actual Anthropic or OpenAI invoice. Treat them as a close, private approximation — and use Token Forest's dashboard to watch your own real numbers over time.

Which one should you pick?

• ✦Pick ccusage if you live in the terminal, want scriptable output for CI, or just need a quick number now and then.
• ✦Pick Token Forest if you want a visual, always-on view, a clickable dashboard, and a bit of gentle motivation.
• ✦Not sure? They don't conflict — install both and use whichever fits the moment.

Can I run both?

Yes. They read the same read-only logs and never modify them, so ccusage in your terminal and Token Forest on your desktop coexist happily. Many people keep ccusage for quick CI checks and Token Forest for the daily at-a-glance picture.