How much does Claude Code cost per developer, and how do I measure it?

Rius measures Claude Code cost per session and per developer by pricing each session’s tokens at API list rates, then totalling them on the Users page. The figure is a list-price estimate in US dollars, so it shows what a developer’s usage would cost at API rates. On a Claude Pro or Max subscription it measures usage and is not what you pay. Rius cannot give a fixed per-developer number in advance, because the cost depends on the models, the tokens and the subagents each developer’s sessions use.

How it works

The Rius plugin turns each Claude Code session into one trace. Every model call in it carries the model, input and output tokens, cache reads and cache writes, and the cost Rius computes from them. You never send a cost value.

• Pricing. Rius prices each call from the model and its token counts, using LiteLLM’s public model-price catalog in US dollars. Cache reads and writes are priced at their own rates. These are list-price estimates, not billed amounts. Discounts, batch pricing and long-context tiers are not applied.
• Subagents. Each subagent gets its own spans with its own tokens and cost. In long sessions, subagents can account for more than half of all tokens, so a session’s total includes them.
• Per developer. With plugin 0.6.2 or later, each span carries the email the developer signed in with as user.id. That is how a session lands under a developer.
• When there is no price. If a model is not in the catalog, or a call reports no token usage, its cost stays empty and never shows as $0. A total with nothing priced in it shows a dash in the console.
• No usage at all. Claude Code reports tokens, so its sessions carry cost. Cursor (beta) does not report usage through its hooks, so Cursor sessions show no tokens and no cost.

The full pricing rules are in Track LLM and agent cost.

Set it up

1. Each developer installs the plugin in a terminal. The Claude Code plugin page covers installing from inside Claude Code and the requirements.

claude plugin marketplace add glassflow/rius-coding-agents && claude plugin install rius@rius-coding-agents

2. Each developer runs /rius:login, picks the shared team workspace and clicks Connect. Use one workspace for the whole team so cost adds up in one place.
3. In each repo you want measured, run /rius:enable-here. It sends structure only: the span tree, models, tokens, cost and timing, with no prompts or replies. Tracing is opt-in per developer and per folder.
4. Run /rius:status and check that the first line reads Rius tracing: on. Start Claude Code from inside the enabled repo, because a session started from a parent folder can stop being traced partway through.
5. Work as usual, then open the Users page in the console. Sort by Cost to see who drives spend, and open a developer to see their recent and failed sessions. The default window is 7 days.
6. To see what one session cost, open it in the trace list. The waterfall shows tokens and cost on each model call, and the Sent by column filters to one developer.
7. To ask from the terminal, run /mcp, pick rius and sign in once. Then ask Claude “Which Claude Code sessions in the last 7 days cost the most, and what did the tokens go on?”

The team rollout, including who can see what, is in Track Claude Code usage and cost across a team. The wider set of coding-agent guides is under Coding agents.

What you get

• Cost per session. One trace per session, with the cost on every model call and the session’s total in the trace header.
• Cost per developer. The Users page lists each developer with their traces, failed traces and cost for the window, sortable by cost.
• Subagent visibility. You can see how much of a session’s cost came from subagents compared with the main session.
• A usage figure on a subscription. If your team is on Claude Pro or Max, the cost shows how much each developer uses Claude Code in API terms. It is not an invoice and not what the subscription costs you.
• Clear gaps. Calls without a price show an empty cost, so a missing price is never mistaken for a free call.

The Users and Agents figures count a session once it has ended. The trace list shows it while it runs. A session that never ends cleanly, because Claude Code crashed or was killed, is never counted there. Sessions from plugin versions before 0.6.2, and sessions sent with an existing API key instead of /rius:login, carry no developer and land in the anonymous row.

When something else fits

• If you need the amount your organization is actually billed, use your provider’s or subscription’s own billing records, because Rius shows list-price estimates and applies no negotiated discounts.
• If your developers use Cursor (beta), Rius shows their sessions but no tokens or cost, so you cannot measure cost per developer there.

Related