Windows
Linux
Custom menu bar & tray
Build a live display from tool icons, quota bars, reset times, cost, tokens, and text, with an instant preview.
Token Monitor
See usage, cost, and limits across every AI coding tool at a glance.
29+ tools and providers
Claude Code
Codex
OpenCode
Hermes
OpenClaw
Cursor
Antigravity
Cline
Kimi
Qwen
Grok
Copilot
Pi
Zed
Kilo Code
MiMo Code
ZCode
Kiro
DeepSeek
MiniMax
CodeBuddy
WorkBuddy
Reasonix
DeepSeek Harness
OpenRouter
Volcengine
One widget, nine views
Every signal is here. Open only the view you need.
Walk through four of the nine real Token Monitor views, or choose a view directly.
TOTAL TOKENS
165,164,772
$136.08
gpt-5.6-sol
55,820,110$46.01
claude-fable-5
38,906,795$32.07
kimi-k3
26,399,329$21.75
glm-5.2
18,975,927$15.62
claude-sonnet-5
14,062,457$11.58
codex-auto-review
11,000,154$9.05
A year of AI coding, charted.
Turn on opt-in history and Token Monitor opens a full dashboard window: a GitHub-style activity heatmap with streaks, plus per-tool and per-model usage stacked over time in bar and K-line views, rolled up across all your devices.
38.42BTotal tokens
$31.86KTotal cost
312Active days
97Current streak
3,101hActive time
430.2MPeak day
gpt-5.6-solTop model
283.0KMessages
Token Activity
By tool
Codex24.96B65.0%
Claude Code9.84B25.6%
Hermes2.74B7.1%
Cursor880.0M2.3%
By model
gpt-5.6-sol15.80B41.1%
claude-fable-510.92B28.4%
kimi-k36.84B17.8%
glm-5.24.86B12.7%
The same usage, on every surface you already use.
Floating Bubble
Collapse the widget into a draggable mini-window with click or hover preview.
Discord Rich Presence
Broadcast today's tokens, cost, and top tool to your profile. Opt-in.
iOS widget
Today's totals on your Home Screen via the Worker hub, with Widgy or Scriptable.
Start with one widget. Add a hub for multi-device sync.
Local stays the default path. Add self-hosted sync when you want token usage from multiple devices rolled into one view.
Local mode
The widget reads local usage summaries through tokscale and renders them on the same machine. No account, no cloud.
There is no mode toggle. Paste a hub URL and the widget starts syncing; clear the field and everything stays on this machine.
Sync mode
Each widget or headless agent posts that device's usage summary to your hub, which merges totals and streams them back to every connected widget.
Pick a sync backend; all three speak the same ingest protocol.
Your code and conversations are not the product.
Token Monitor syncs only the fields needed to show totals, costs, tool and model breakdowns, and normalized account limit status.
Never syncs
- Raw prompts or source files
- Conversation transcripts
- OAuth credentials or provider responses
{
"deviceId": "mac-studio",
"hostname": "Mac Studio",
"platform": "darwin",
"agentVersion": "0.36.0",
"updatedAt": "2026-07-28T10:14:20.193Z",
"today": {
"totalTokens": 165164772,
"costUsd": 136.08,
"clients": {
"codex": 98123456,
"claude": 49728853,
"kimi": 10000000,
"zai": 7312463
},
"models": {
"gpt-5.6-sol": 67882721,
"claude-fable-5": 46906795,
"kimi-k3": 29399329,
"glm-5.2": 20975927
}
},
"month": { … totals, clients, models … },
"allTime": { … totals, clients, models … },
"limits": {
"providers": [
{
"provider": "claude",
"accountKey": "sha256:9f2c…",
"status": "ok",
"windows": [{
"kind": "weekly",
"usedPercent": 61.4,
"resetsAt": "2026-06-15T00:00:00Z"
}]
}
]
}
}The entire record a hub ever receives: counts, costs, and labels. Nothing else fits in the shape.