Quick answer
The local whoburnedmore CLI command is npx whoburnedmore --local. Decline its optional publish offer to keep the report local. The standard command signs in and submits daily aggregates. Neither mode submits prompts, source code, or filenames.
What this guide answers
The reader is choosing between tools or trying to explain why spend moved from one to the other.
Best for
Developers, team leads, and founders deciding where to run agentic coding work.
Key facts, checked November 8, 2026
- Claude Code comes with every paid Claude plan: Pro at $20/month ($17/month billed annually), Max at $100/month (5x Pro usage), and Ultra at $200/month (10x Pro weekly usage, no 5-hour limit). You can also pay API rates through a Console account. source
- Anthropic's figure for Claude Code on enterprise deployments is about $13 per developer per active day and $150-250 per developer per month, with 90% of users below $30 per active day. source
- New Cursor plans are sold as the X Subscription: Plus $8, Premium $30, Super $100, Ultra $200, and Teams at $30, $100 or $200 per seat. The June 2026 Teams update ($40 Standard and $120 Premium seats) now applies only to legacy Teams plans. source
- Cursor plans include a weekly pool for Cursor's own models. Usage beyond the pool is billed per million tokens (Composer 2.5 is $0.50 in and $2.50 out), and input over 256K tokens of context costs 2x. source
- On Cursor Premium and above, third-party models are billed at API list price from prepaid credit. Teams and Enterprise pay an extra $0.25 per million tokens. source
Compare the same work
A fair comparison needs the same task, files, and acceptance test. Otherwise you are comparing a debugging rabbit hole in one tool to a small edit in another.
- Use the same repository.
- Define done before starting.
- Record retries and rewrites.
Know where each tool shines
Cursor is strongest when IDE context and inline workflows matter. Claude Code is strongest when terminal automation and repo-wide actions matter.
- IDE edits
- Terminal workflows
- Long-running agents
- Review loops
Account for visibility
Cursor has been adding usage visibility for teams, while terminal workflows often need a separate tracker. Visibility matters because hidden cost becomes culture debt.
- Show spend by user.
- Show spend by project.
- Warn before large loops.
Measure your own mix
Most teams will keep both tools. The smart move is to route work: IDE-native edits to Cursor, terminal-heavy repo tasks to Claude Code, and all usage into one tracker.
- Route by task type.
- Review weekly cost drift.
- Keep one shared leaderboard.
The billing units are different, so compare on the same task
Claude Code on a subscription spends a shared allowance. It has a 5-hour session limit plus a weekly limit on Pro and Max, and only a weekly limit on Ultra. Cursor spends a weekly model pool first, then per-token on-demand usage. Running out therefore looks different in each tool. In Claude Code you wait for a reset or buy usage credits. In Cursor you start paying per token, up to whatever additional-usage limit you set. Writing down 'Claude Code cost $X' and 'Cursor cost $Y' only means something if both numbers come from the same task.
- Claude Code: run /usage for the session cost estimate and your plan usage bars.
- Cursor: open the usage page in the dashboard and note which pool was used, or whether on-demand usage kicked in.
- Log the model in both tools. A frontier model in one and a small model in the other is a model comparison, not a tool comparison.
Where the money actually goes
Anthropic's docs say Claude Code resends the whole conversation with every request, and each tool call adds another request that carries its results. Long sessions and big file reads compound. Cursor has its own lever: input over 256K tokens of context costs 2x on Cursor-model usage. In both tools the cheapest habit is the same. Clear or start a new session between unrelated tasks, and keep huge logs out of the context.
- Claude Code: /clear between tasks, /compact with a focus, Sonnet or Haiku for routine work.
- Cursor: keep long agent threads short, and check whether Auto or a premium model handled the request.
Track both in one place
whoburnedmore reads Claude Code's local logs and Cursor's usage through Cursor's own usage API, along with 20 other agents. Its daily view shows which tool burned what on the same days. It reports token totals and estimated cost, not your remaining Cursor pool or Claude plan limit. For those, use /usage in Claude Code and the Cursor dashboard.
Common questions
Is Claude Code cheaper than Cursor?
It depends on the plan and the work. Claude Pro is $20/month with session and weekly limits. New Cursor plans start at $8 (Plus) and $30 (Premium) and add per-token charges past the pool. Measure the same task in both before deciding.
How do I see my Claude Code usage?
Run /usage inside Claude Code. It shows the session cost estimate and, on subscription plans, your plan usage bars. /cost and /stats are aliases for it.
What happens when I run out of Cursor's included usage?
Cursor-model usage is then billed per million tokens, and third-party models draw on prepaid credit at API list price. Auto Top-up and a monthly additional-usage limit control how far it goes.
Can one tracker show Claude Code and Cursor together?
Yes. npx whoburnedmore reads Claude Code's local logs and Cursor's usage API, plus 20 other supported agents, and groups them by day, tool and model.