Usage, cost, tool limits and subscriptions sync automatically.
Optional content sync is available when connected to or hosting a hub.
Title sync is off, but previously shared titles still need to be removed from the server. Retry when connected.
Token MonitorΣToken Monitor
Starting
Startup
App UpdatesGitHub releases
Installed—
LatestNot checked
Integrations
TokscaleBundled CLI dependency
InstalledChecking...
Bundled build
Token Monitor uses tokscale (MIT, open-source) to read token data.
Additional options are available in settings.json.
About Token Monitor—
Open-source AI tool token usage monitor, licensed under MIT.
Diagnostics & support
Create a redacted system diagnostic to help report issues.
Includes versions, collection state and token totals; excludes credentials, conversations, accounts, full paths and raw errors.
Report preview
Choose which main views appear and the order of the mode button.
Exchange rate
1 USD =
Model ranking
Shortcut
Record a global shortcut to show or hide the window.
Floating & Tray
A slim quota rail hidden at the screen edge. Push the pointer to the edge to reveal it, click to pin, drag to move or switch sides.
Visibility
Screen edge
Size
Custom size
Collapse the widget into a slim draggable mini-window; click it to reopen.
Open on
Hide the main window; show a popover from the macOS menu bar or Windows system tray.
The widget stays on the desktop; use the tray icon or the shortcut to bring it to the front.
Glass Backdrop
GlassManaged by macOS
DepthManaged by macOS
Zoom
Background image
No image selected
Image opacity
Reduce Motion
Reduce animations, or follow the system setting.
Interface Theme
Shares interface colours only. Vendor colours and appearance preferences stay local.
Text sizeAdjust labels and data text. Token total and icons keep their size.
Preview
Aa 中文 0123
Preview
Aa 中文 0123
App default uses Token Monitor's built-in typography. System UI uses your operating system's interface font. Choose Custom… to enter a font family.
Override the chart and list colour for each tool. Reset returns it to the brand colour.
Track controls collection. Show controls the main Tools list. Drag a row to reorder.
Session history
Data export
Export your usage as CSV + JSON for spreadsheets, Obsidian, Grafana, or scripts.
Or export once to a folder you pick:
Show different names for the same model as one row.
Automatic grouping
Merge duplicates only handles a model reported under two spellings at once. Strip prefixes removes the provider name before the / in every model name, for example anthropic/claude-opus-5 → claude-opus-5.
All aliases above will point to this model ID.
Override a model's price (USD per 1M tokens) when its auto-detected price is wrong. Applies to all providers of that model.
Rates · USD / 1M tokens
Cache-write pricing
Leave blank to use the regular cache-write price.
If the supplied rates cannot price the usage, the original cost is kept.
Pricing guide
Unused rates can be left blank. Custom pricing applies only when the rates cover every used token type.
For a free model, set both input and output to 0 and leave other prices blank or 0.
Cache-write rates are only needed for providers that bill cache writes separately. If the 1-hour rate differs, enter it manually.
Drag a provider to reorder the list.
Adjusts the refresh rate to how fast a quota is being consumed: every 5 minutes normally, as often as every minute when needed.
Bars show
Sign in to each Codex account you want to track. Tracking uses isolated managed logins; switch the local Codex account from the Limits page.
Details
Go quota is detected automatically. Save a credential under an existing account's name to say they are the same OpenCode account: quota then comes from the API, with the Zen balance and cross-device identity still coming from the cookie.
Paste an OpenCode Go API key. Go quota only.
1. Open opencode.ai/auth in your browser and sign in.
2. Open DevTools (F12 or Cmd+Opt+I) -> Application/Storage -> Cookies -> opencode.ai.
3. Copy the auth value.
4. Paste it below and click Save.
Token Monitor automatically detects accounts signed in through the Cursor desktop app. Uncheck an account to stop monitoring its limits; account-level token usage still includes every saved account.
Stop the headless agent before adding or removing Cursor accounts.
1. Open cursor.com/dashboard in your browser using the button above, then sign in.
2. Open DevTools (F12 or Cmd+Opt+I) -> Application/Storage -> Cookies -> cursor.com.
3. Copy the WorkosCursorSessionToken value.
4. Paste it below and click Save.
Paste your Kimi Code API key. It stays on this device and is used only for the Kimi Code quota API.
Optional: Kimi Web fallback
Use a Kimi Web access token only to fill quota windows missing from the Code API. The token stays on this device and expires quickly.
1. Open Kimi Code Console in your browser (button above) and sign in.
2. Open DevTools (F12 or Cmd+Opt+I) -> Application/Storage -> Local Storage -> https://www.kimi.com.
3. Find access_token and copy its Value.
4. Paste the value below and click Save Web token.
Paste a GitHub OAuth token with Copilot access. It stays on this device and is used only for Copilot quota lookup.
No accounts yet. Click "+ Add account" below to get started.
1. Open the Xiaomi MiMo console in your browser (button above) and sign in.
2. Open DevTools (F12 or Cmd+Opt+I) -> Network, then reload the page.
3. In Network, select balance, then copy the full Cookie value from Headers -> Request Headers.
4. Paste it below, then click Save account.
Add an OpenRouter API key to view usage, the key limit, and available balance. Both regular and Management API keys are supported. Keys stay on this device and only official OpenRouter endpoints are used.
Paste an Ark API key, or use a Volcengine access key ID with the secret field. An access key reads both the Coding Plan and the Agent Plan.
Leave empty to read the Agent Plan with the access key above. Fill these in only when the Agent Plan was bought on a different Volcengine account.
Connect a compatible relay or balance API. Credentials stay on this device and are sent only to the Base URL below.
Shows the whole account balance. New API requires a User ID; compatible One API forks can leave it blank.
HTTP sends credentials without transport encryption. Use HTTPS unless you trust the network.
1. Open the configured Sub2API Base URL in your browser and sign in.
2. Open DevTools → Application → Local storage, then select the dashboard origin.
3. Copy the value of auth_token below. You can also copy refresh_token for automatic renewal.
4. Save the account. Prefer a dedicated dashboard session because renewing refresh_token may sign the source browser out.
Response mapping
Dot-separated JSON paths
Record what you actually pay for each AI account. Hover an account's plan label on the AI Tool Limits page to see what you recorded here.
Record
The date money first left your account. A free trial start does not count.
Leave empty to work it out from the first charge. Set it only when a trial, plan change or promotion shifted the date once.
Top-ups
Record every top-up. This month's total is what gets compared against this month's usage.
Other devices connect using one of the URLs above with the shared secret. macOS / Windows may prompt to allow incoming connections the first time. For cross-network use, run Tailscale or ZeroTier on both sides - the virtual IP works the same.
The hub runs while Token Monitor is running. Quitting (not just closing the window) stops the hub for all connected devices.
Sync usage, limits, history, devices and subscriptions across Macs signed into the same Apple ID. No Token Monitor server is used; updates may take a moment to appear.
Devices
TOTAL TOKENS
0
$0.00