Skip to content
Menu

Guides

ccusage, tokscale, UsageAtlas, CodeBurn and the other local trackers vs OpenReason Intel

Checked on 4 November 2026 against the sources below. Updated on 12 November 2026.

The local trackers and OpenReason Intel answer different questions. ccusage, tokscale, UsageAtlas, CodeBurn, OpenUsage, Usagebar, claude-session-budget, ClaudeCapper and Claude Code Usage Monitor each run on one person's machine, read what the coding tools already store there, need no sign-up, and are free or nearly free. Several of them notify you as a limit gets close, and three can stop Claude Code before a limit or a budget is reached, which OpenReason Intel does not do. OpenReason Intel is a dashboard for Claude Code and Codex usage, for one person or a whole team. One command on each machine sends usage to the dashboard, where you can see your 5-hour and weekly history from every machine you use, the turns that filled each window, and what you spent past your plan; a lead sees the same for every engineer, with seats, spending and recorded pull requests alongside.

If you are one engineer and all you want is to see your window while you work, a free local tracker does that well, and you may not need anything more. A menu bar app or a status line puts the percentage in front of you, and some will stop Claude Code before the limit. If you want to see what used up your window and keep that record, OpenReason Intel adds four things for one person: Claude Code and Codex in one place; a record across time and across every machine you use, where each local tracker sees the machine it runs on; each turn's estimated share of the window, so you can see which work filled it; and a record you can show your lead when you ask for a bigger seat or more budget. It is free for one person. If you pay for a team, the local trackers cannot show you who used what, because each one sees a single machine. Every row below is taken from the tool's own site or README, read on 4 November 2026.

The comparison, from each tool's own pages

ToolWhat it reads and showsCoding tools it coversDoes it notify or stop anything?Where it runs, and the price
ccusageIt reads the usage logs coding agents write on your machine and produces daily, weekly, monthly, session and 5-hour block reports with costs at API list prices. A block's percentage is measured against your own busiest past block, not your plan's limit, and it has no weekly limit.It covers 19 agents, including Claude Code, Codex, OpenCode, Gemini CLI and GitHub Copilot CLI, but not Cursor.It does not notify or stop anything; you can set a token limit of your own to mark reports against.It is a terminal command with no account, free under the MIT licence.
tokscaleIt reads local session files for token and cost reports at API prices, and asks Anthropic and OpenAI, with the sign-in already on your machine, for your current 5-hour and weekly percentages. It shows them as they are now and does not keep them over time.It covers 59 tools, including Claude Code, Codex, Cursor, OpenCode, Gemini CLI and GitHub Copilot CLI.It does not notify or stop anything.It is a terminal tool with no account, free under the MIT licence. If you sign in with GitHub and choose to submit, your daily totals go to a public leaderboard.
UsageAtlasIt reads local activity and reuses each tool's sign-in to show token history, cost estimates, credits and usage limits over 7, 30 and 90 days.It covers Codex, Claude Code, Cursor, OpenCode, GitHub Copilot and about ten others, some with balances only.It sends one desktop notification per reset cycle when a window passes the threshold you choose; it does not stop anything.It is a desktop app for Mac, Windows and Linux with no account needed, free under the MIT licence, and labelled beta.
CodeBurnIt reads local session files and asks each provider how much of your plan is left, showing cost by model, project and session, a 5-hour and weekly row for Claude, and cost per pull request.It covers Claude Code, Cursor, Codex and 39 other tools.Its opt-in Guard feature gives a one-time warning at a soft dollar cap and stops a Claude Code session at a hard cap.It runs as a Mac menu bar app, a Windows tray app, a desktop app, a VS Code extension or a command, with no account and no paid tier, under the MIT licence.
OpenUsageIt reads local logs and the stored sign-in to show live quota, spend, rate limits, burn rate and reports, kept in a local database.It covers Claude Code, Codex CLI, Cursor, GitHub Copilot, Gemini CLI, OpenCode and other tools and API platforms, 35 in all.It can flash a tmux banner or ring a bell at thresholds you set, which are off by default; it does not stop anything.It is a terminal dashboard for Mac, Linux and Windows with no account, under the MIT licence.
UsagebarIt uses the Claude Code sign-in stored in the macOS Keychain to ask Anthropic for your 5-hour and weekly limits and shows them with reset timers.It covers Claude Code only.It notifies you at 50, 75 and 90 per cent of a limit and when your session context runs low; it does not stop anything.It is a macOS menu bar app that costs $9 once, with lifetime updates.
claude-session-budgetIt reads Claude Code's local session files to estimate how much of the 5-hour window you have used.It covers Claude Code only, and the 5-hour window only.At 93 per cent of the window it blocks tool calls by default, and it can instead wait and re-check.It is a Claude Code hook installed from the plugin marketplace, pip or Homebrew on any system, free under the MIT licence.
ClaudeCapperIt uses Claude Code's stored sign-in to ask Anthropic for the 5-hour, weekly and per-model limits and shows them as live bars.It covers Claude Code only.It blocks Claude Code when any limit reaches a threshold, 90 per cent by default, "so no extra usage credits are spent".It is a Windows tray app with PowerShell hooks, free under the Apache 2.0 licence.
Claude Code Usage MonitorIt reads Claude Code's local logs or the status line's limit fields to show the 5-hour window, burn rate, cost and a forecast of when you will run out.It covers Claude Code only.It gives predictions and warnings as a limit gets close; it does not stop anything.It is a terminal tool for Mac, Linux and Windows, local only, free under the MIT licence.
Session Forensics (OpenReason)It reads your Claude Code and Codex session files in your browser and lists the limits you hit in the last seven days, with the turns that used each window up and their estimated share. For Claude Code the 5-hour figure is an estimate, and the weekly window is not drawn.It covers Claude Code and Codex.It does not notify or stop anything.It is a web page with no install and no account, and your files are never uploaded. It is free.
OpenReason IntelA reporter on each machine sends Claude Code and Codex usage and limits to a dashboard showing a person's window history across machines, the turns that filled each window with their estimated share, periods past the limit, spending and, for a team, recorded work.It covers Claude Code and Codex; OpenCode support is coming.It does not notify or stop anything; a team with a coding agent can build its own warning on the data.It runs on Mac, Windows and Linux with an account, free for one person and for teams under 10 engineers, and $20 an engineer a month above that.

Sources for each row are listed at the bottom of the page. Small open-source projects change quickly, so check a tool's own page before you rely on a detail.

Where the local trackers do more

  • They are free and need nothing from anyone. Eight of the nine are open source, Usagebar costs $9 once, and none of them asks you to create an account. Your usage stays on your machine, apart from the calls some of them make to the vendor to read your limits.
  • Some of them act, which OpenReason Intel does not. claude-session-budget blocks tool calls at 93 per cent of the 5-hour window, ClaudeCapper blocks Claude Code at a threshold so that usage credits are not spent, and CodeBurn's Guard can stop a Claude Code session at a dollar cap. UsageAtlas, OpenUsage, Usagebar and Claude Code Usage Monitor notify you as a limit gets close.
  • Several cover more tools. ccusage, tokscale, UsageAtlas, CodeBurn and OpenUsage each read a dozen or more coding agents, including Cursor in the case of tokscale, UsageAtlas, CodeBurn and OpenUsage. OpenReason Intel reads Claude Code and Codex.
  • They show the live window where you are working, in the menu bar, the tray, the status line or a terminal pane, which is where an engineer looks while a session runs. Several, among them tokscale and Usagebar, ask the vendor for your current percentages, so the figure is the vendor's own.

When a local tracker is enough

Plainly: if you use Claude Code on one machine, you only want to know how close you are right now, and you have no reason to show anyone else, a local tracker or Claude Code's own status line is enough. Usagebar, UsageAtlas or Claude Code Usage Monitor will keep the figure in view, and claude-session-budget or ClaudeCapper will stop Claude Code before the limit if that is what you want. OpenReason Intel does neither of those things.

What OpenReason Intel adds for one person

  • A record across machines. Claude Code's own /usage figures come from "local session history on this machine, so usage from other devices or claude.ai is not included", and every local tracker in the table reads the machine it runs on. If you work on a laptop and a desktop, or a work machine and a home one, OpenReason Intel puts them in one record.
  • What filled each window. When a 5-hour or weekly window filled, OpenReason Intel shows the sessions and turns that were running, each with its estimated share of the window, so "what used it all up?" has an answer. The local trackers report tokens, cost and percentages by day, session or 5-hour block; none of their own pages, read on 4 November, describes tying a filled window to the individual turns that filled it, and tokscale's, read on 12 November, does not either. OpenReason's free Session Forensics does this for the last seven days on one computer, without an account; Intel keeps it for every window on every machine.
  • Claude Code and Codex in one place, over time. ccusage, UsageAtlas, CodeBurn and OpenUsage also read both tools and keep a history on one machine, so this is not Intel's alone; what differs is that Intel's record is the cross-machine one above.
  • A record you can show someone. If your company pays for your seat and you need a bigger one or more budget, a link to your own window history and what filled it is easier to argue from than a screenshot of a menu bar. The guide on asking your company for more budget goes through it.

What OpenReason Intel adds for a team

  • It shows a whole team. Every local tracker reads one machine and shows one person. A lead who wants to know which engineers filled their window every day last week, which seats sat idle and who used paid credits would have to ask each engineer to run a tool and report back. OpenReason Intel collects that from every machine into one place.
  • It shows Claude Code and Codex together, with seats and variable spending in one Spend view, so a lead does not have to open two vendor consoles.
  • It puts usage beside recorded work. The Work page shows branch and feature activity beside agent time, window use and recorded pull request links. CodeBurn shows a cost per pull request on one machine; the others measure tokens rather than what the tokens produced. A missing pull request does not mean no useful work was done, and a pull request is not a measure of value.
  • It keeps each person's window history and the periods past the limit, from every machine they use, so a lead can inspect a pattern afterwards rather than ask each engineer to report back.

What it costs you in return, whether you are one person or a team: it needs a sign-up and a team (a team of one is fine), the usage leaves your machine and goes to OpenReason, and by default that includes prompt and reply text, which --no-log-prompts at install turns off. The or-otel quickstart and the engineers' page list exactly what is sent.

Using both

The two kinds of tool do not conflict. An engineer can run Usagebar or claude-session-budget to stay under their own window while OpenReason Intel's reporter sends the same sessions to the team's dashboard. One thing to know: OpenReason Intel sets Claude Code's status line to its own command and then runs the status line you had before, so a status-line tracker keeps working through it.

What to do next

  1. Read next

  2. See what used up your limit

    Session Forensics is a free page that reads the Claude Code and Codex session files already on your computer, in your browser. It lists the limits you hit and the turns that used each one up. There is nothing to install and no sign-up, and your files are never uploaded. For Claude Code its 5-hour figure is an estimate, because Claude Code's files do not record the meter.

    Would you rather have your own agent do it? Paste this prompt into Claude Code or Codex.

    Open Session Forensics (opens in a new tab)

  3. Get an email when the limits change

    Get an email when Claude Code or Codex limits change. One email per change, at most one a week, from OpenReason. Unsubscribe from any email in one click.

  4. Track it over time

    Sign up with your Google account, no card, and run one install command on each machine you use. OpenReason Intel then keeps your Claude Code and Codex windows in one record, with the turns that used each one up. It is free for one person, and it runs beside a local tracker without replacing it.

    It shows you what happened; it does not stop a session as you work, and it cannot change your vendor's limits.

    Sign up free with Google

Questions people ask

ccusage vs OpenReason (and the other local trackers).
ccusage is a free local command that reports tokens and estimated cost from the logs on one machine, for Claude Code, Codex, OpenCode and many other agents, with no sign-up. OpenReason Intel is a hosted dashboard that collects Claude Code and Codex usage from every engineer's machine and shows window history, periods past the limit, seats and variable spending, and recorded work per person. For one person, ccusage is enough to see how much you used on one machine, though its percentages are measured against your busiest past window rather than your plan's limit. OpenReason's free Session Forensics page shows the limits you actually hit in the last seven days and the turns that used each one up, in your browser with nothing uploaded; OpenReason Intel adds a record across machines and every window over time, free. For a lead, only OpenReason Intel shows the whole team.
tokscale vs ccusage vs OpenReason.
tokscale and ccusage are both free terminal tools that read your coding agents' local files and price the tokens at API rates. tokscale reads more tools, 59 against ccusage's 19, and it can also show your current Claude Code and Codex 5-hour and weekly percentages by asking the vendor with your stored sign-in, which ccusage cannot. Neither keeps those percentages over time or shows which turns used up a window. OpenReason's Session Forensics, a free browser page, lists the limits you hit in the last seven days and the turns behind each one; OpenReason Intel keeps your real 5-hour and weekly figures over time from every machine, and shows a team lead the same for every engineer, in private. tokscale's team feature is a leaderboard of tokens and dollars, and what is submitted to tokscale.ai is public; a team that wants it private has to run tokscale's website on its own server.
How do I check how close I am to my Claude Code limit?
Run /usage in Claude Code for the 5-hour and weekly bars. To keep them in view, a local tracker such as Usagebar, UsageAtlas, CodeBurn or Claude Code Usage Monitor shows the percentages on your machine as you work. OpenReason Intel shows the same windows as a history across all your machines, with the turns that filled each one.
How do I track my Claude Code usage over time?
ccusage gives daily, weekly and monthly reports from the logs on your machine, UsageAtlas shows 7, 30 and 90 days, and OpenUsage keeps a local database. OpenReason Intel keeps your window history and the periods past the limit across all your machines, for one person or for every person on a team.
Can I see my Claude Code usage in the status line?
Yes. Claude Code passes your 5-hour and weekly percentages to a custom status line, and Claude Code Usage Monitor can read those fields. ccusage has a status line too, but it shows cost and context rather than those percentages. These are per machine; OpenReason Intel uses the status line to collect the figures and keeps them as a history across your machines.
How do I see what my team spends on Claude Code?
A local tracker cannot show a team, because each one reads one machine. Anthropic's Team spend report shows usage-credit spend per person once usage credits are on. OpenReason Intel shows seats and variable spending together per person, for Claude Code and Codex.

Sources

Every figure on this page was read from these pages on 4 November 2026. Limits and prices change; if a vendor page now says something different, the vendor page is right and this one is out of date.