updated October 3, 2026
How to track Claude Code usage and cost across your team
Anthropic gives you a dashboard for each way of paying: claude.ai/analytics/claude-code on Team and Enterprise (active users, sessions, accepted lines and, with GitHub connected, pull requests) and platform.claude.com/claude-code for API organizations (the same activity, and spend per person). For tokens and cost by person in your own tools, Claude Code sends OpenTelemetry. None of them covers Codex, personal accounts used for work, or who is about to hit a limit.
Your options at a glance
| Option | For | Shows | Leaves out |
|---|---|---|---|
| Claude's analytics dashboard | Team and Enterprise admins and owners | Daily active users, sessions, accepted lines and accept rate; pull requests and lines shipped with GitHub connected; a top-10 leaderboard; CSV export | Dollars for use inside seats, API usage, Codex |
| The spend report | Team and Enterprise | Tokens and estimated usage-credit spend per person and model, as a CSV | Use inside seats, which is not metered in dollars |
| Claude Console dashboard | API organizations | Accepted lines, accept rate, active users and sessions, daily spend, and spend and lines per person this month | Subscription seats, pull requests, Codex |
| OpenTelemetry | Any setup, cloud providers included | Sessions, tokens, cost, lines, commits, pull requests and active time per person, every minute | Somewhere to keep and chart it |
Claude's analytics dashboard, on Team and Enterprise
Admins and owners find it at claude.ai/analytics/claude-code. It counts daily active users and sessions, the lines of code people accepted and how often they accept Claude's edits. With the Claude GitHub app installed and GitHub analytics turned on by an owner, it adds contribution metrics: merged pull requests and lines shipped with Claude Code, matched against Claude Code sessions, and a leaderboard of the top ten contributors, exportable for everyone. Contribution metrics appear within a day of turning them on and update daily; they are in beta and not available with Zero Data Retention.
Seats are a fixed price, so the dashboard has no dollars for what people use inside them. The spend report in the organization's analytics lists tokens and estimated spend per person and model for usage credits, and on Enterprise the Enterprise Analytics API returns per-person usage and cost to your own systems.
The Claude Console, for API organizations
When people sign in to Claude Code with a Console account, their usage goes to a workspace called Claude Code, created for it. The dashboard at platform.claude.com/claude-code, open to the Developer, Billing, Admin and Owner roles, shows accepted lines, the accept rate, daily active users and sessions, daily spend, and a table of each member's spend and lines this month. The Claude Code Analytics API returns the same daily numbers per person with an Admin API key. Its spend is an estimate; the billing page has the invoice.
OpenTelemetry, for tokens and cost per person
Claude Code can send its own metrics and events to any OpenTelemetry collector: sessions, tokens, cost, lines changed, commits, pull requests and active time, by person, whichever way they pay, including Amazon Bedrock and Google Cloud, which the dashboards above do not see. It is the only option Anthropic describes as streaming per-person tokens and cost in near real time. What it does not give you is the dashboard: the numbers land in Datadog, Grafana or whatever you run, and the charts, the people and the prices are yours to build.
What none of them show
- Codex. OpenAI's tools have their own console, so a team on both has two places to look and two ways of counting.
- Personal accounts used for work. A developer's own Pro or Max is not in the company's organization, so its admin dashboards never see it.
- Limits. How close each person's account is to its 5-hour and weekly limits, who keeps running out, and when it comes back.
- Money your way. Spend by team, project and cost center, seats at the rates you actually pay, budgets, forecasts and chargeback, and Cursor and GitHub Copilot beside Claude.
Claude Code and Codex in one place
Tokaware puts all of it on one console: every Claude Code and Codex session as it runs, with its person, project and computer; tokens, prompts and lines by person, project and model; every account's limits; and spend from seats and from what Anthropic, OpenAI, Cursor and GitHub bill. The numbers come from a desktop app on each computer, from Claude Code's and Codex's own OpenTelemetry on Business, and from the providers' admin APIs, which bring in the last 30 days with nothing to install.
Computers share numbers only unless the organization's policy and each developer allow more. It is free for up to 3 people, $5 a person a month on Team, and new teams get 14 days of Business without a card.
for your team
See your team's Claude Code and Codex in one place
Tokaware shows every session, limit and dollar across your developers, by person, team and project. Free for up to 3 people, and new teams get 14 days of Business, no card needed.
Questions people ask
- Can I see Claude Code usage per developer?
- Yes. API organizations see spend and lines per member in the Claude Console. Team organizations export the spend report as a CSV, and Enterprise organizations can also use the Enterprise Analytics API. OpenTelemetry gives tokens and cost per person on any setup.
- Does Claude's analytics dashboard show cost?
- Not for use inside seats, which cost the same however much they are used. The spend report lists estimated spend of usage credits per person and model, and the Claude Console shows daily API spend.
- Who can see the Claude Code analytics dashboard?
- On Team and Enterprise, admins and owners. In the Claude Console, the roles with the UsageView permission: Developer, Billing, Admin, Owner and Primary Owner.
- Does it count Claude Code on Amazon Bedrock or Google Cloud?
- No. Claude Code sends nothing from cloud providers back to Anthropic, so the dashboards and the Analytics API do not cover it. OpenTelemetry or an LLM gateway does.