tokenburn
Local-firstPrivate by default

Every token you burn, in one place

See how you really build with AI. Every model, token and streak across your tools, on one dashboard and one @profile worth sharing.

$npm install -g use-tokenburn && tokenburn setup
tokenburn.site/dashboard
OverviewSynced 4 min ago
OverviewYour AI coding usage · last 30 days · UTC
Total tokens
597M
Across 3 tools
Estimated cost
$735
API list prices, not a bill
Requests
11K
Across all tools
Current streak
26 days
Longest 26 days
Output tokens per day
Stacked by model
Claude Opus 5.5 · Claude Codegpt-6-astra · Codex CLIgemini-3.8-flash · Gemini CLI
0500K1M1.5MOct 2Oct 5Oct 8Oct 11Oct 14Oct 17Oct 20Oct 23Oct 26Oct 29

Works with the tools you already use

Your profile

Show the work behind the commits

A shareable page for the models you run, the tokens you burn, and the streaks you keep. Flip a switch, preview live, then publish.

What your profile shows

Same switches as Settings. Private until you publish.

tokenburn.site/@yourname
tokenburn
YO

yourname

@yournameyourname
Since Aug 30Claude CodeCodex CLIGemini CLI
Output tokens per day
Since Oct 2 · UTC days
Claude Opus 5.5 · Claude Codegpt-6-astra · Codex CLIgemini-3.8-flash · Gemini CLI
0200K400K600KOct 2Oct 5Oct 8Oct 11Oct 14Oct 17Oct 20Oct 23Oct 26Oct 29
Last 90 days
665.1M tokens
Best day · Oct 30
Aug 27Today
Current streak
41 days
Aug 27Today
By day of week
Fridays
MonTueWedThuFriSatSun
Contributions
105 active days in the last 6 months
LessMore
Models
Last 90 days · select one for details
Sort by
Features

Everything you'd want to know, nothing you wouldn't share

One place for every AI coding tool, built for developers who live in the terminal.

Activity

Watch the tokens fly

Tokens, requests and cost stacked by day and by model, so you see what actually moved. Your biggest day gets its own spotlight.

TokensRequestsCost
30 days agoToday
Streaks

Don't break the chain

Any tool, any day keeps it alive. Then go chase your longest run.

30 days agoToday
Heatmap

Your year, square by square

A GitHub-style heatmap of every day you built with AI, across every tool you use, all year long.

Models

Know your heavy lifters

See which models actually do the work. Rank by tokens, requests or cost, then dive into any one.

Claude Opus 5.552%
GPT-6 Astra28%
Gemini 3.8 Flash12%
Local-first

Your code stays home

A tiny CLI reads usage logs right on your machine. Prompts and code never leave it.

➜tokenburn sync·✔ 3 tools found
Auto sync

Set it and forget it

A quick sync every 30 minutes, run by your OS. Nothing humming in the background.

every 30 min
Costs

No mystery math

Billed where providers report it, list price where they don't. Always labeled.

BilledTool estimateList price
How it works

Less setup than a coffee

Install, set up, then code as usual. Pick a step to watch it run in the terminal.

~ zsh
➜ ~
Privacy

Your code stays on your machine

tokenburn only ever sees totals. Here's where your data goes, step by step.

Never leaves your machine
Prompts and responsesSource code and file pathsProject and machine namesAPI keys and credentials
Synced: only totals
Tokens per model, per dayRequest countsHashed session idsCursor and OpenRouter rows
On your profile: your call
Private until you publishA switch for every fieldPreview before going publicDelete everything in Settings

Setup shows the exact upload and asks before the first one.

Questions, answered

Are my prompts or code uploaded?

No. The CLI reads usage records on your machine and uploads daily totals per model, request counts and hashed session ids. Prompts, responses, code, file paths, project names and keys stay on your computer.

Is my profile public?

Not until you say so. Your profile is private by default; publish it from Settings or with tokenburn profile public, and pick exactly which fields it shows.

What's on a public profile?

Only what you switch on: your tools, models, token totals, a daily chart, streaks, a contribution graph and estimated cost. Each one has its own switch, and the preview above uses the same ones.

How is cost estimated?

The provider's billed amount where reported (OpenRouter), otherwise the tool's own estimate, otherwise public API list prices. It is not what a subscription costs, and it's always labeled as an estimate.

Does something keep running in the background?

No. Your OS runs a quick sync every 30 minutes and nothing stays running in between.

How do I sync right now?

Run tokenburn sync in your terminal. It collects usage on this machine and uploads it immediately. After setup, auto sync does the same every 30 minutes on its own.

How do I get help with the CLI?

Run tokenburn help or tokenburn --help for the full command list. Bare tokenburn prints the same thing: stats, sync, profile, service, and the rest.

Does Cursor work with tokenburn?

Yes. Export a CSV from cursor.com/dashboard → Usage, then drop it in your Downloads folder. With auto-import on (in setup or tokenburn cursor auto-import on), every sync picks it up. Download a fresh export about once a month. You can also import one file with tokenburn import-cursor <file.csv>.

Can I delete everything?

Yes. Deleting your account in Settings removes your profile and all uploaded usage. tokenburn forget clears the local data on each machine.

You've burned millions of tokens.Got anything to show for it?

Claim your handle and turn every session in Claude Code, Codex, Cursor and Gemini into streaks, stats and a profile worth sharing.

$npm install -g use-tokenburn && tokenburn setup
tokenburn

Every token you burn, in one place. Collected on your machine, shared only if you choose.

Local-first · Private by default
© 2026 tokenburnUsage is collected locally · costs are API list-price estimates