Claude Code · Pro / Max / API
Your usage.
The price comparison.
Import a month of Claude Code usage, compare its API-equivalent value with your plan price, and export a receipt. Try the example first. Your report stays in this browser.
Monthly US web prices before tax. API value measures recorded tokens, not remaining quota. Cache-write lifetime is a scenario choice because ccusage's aggregate totals do not identify it.
Run npx ccusage@latest claude monthly --json on the computer where you use Claude Code.
Your usage stays here. Calculations and image export run in this browser. We count anonymous actions, never the report or its results.
Example data. These are made-up numbers, not a customer's usage. Clear them before importing yours.
The same recorded tokens, next to each price.
| Plan | Price | Break-even | Your multiple | Cost comparison |
|---|
Month by month, priced at API rates.
| Month | Claude Code | vs plan | Other |
|---|
Claude Code rows are attributed from the report's agent field, or inferred from unprefixed Claude model IDs in older exports. Other includes other agents and unclassified usage. The selected plan price is applied to every month for comparison; it is not your billing history.
Your comparison, ready to share.
The card shows API-equivalent value against the selected monthly price. It includes the pricing date and cache-write assumption. It does not show subscription capacity or actual savings.
Paste your usage above to generate yours
No usage data? Estimate instead.
Pick a workload to estimate API-equivalent monthly cost. These presets cannot predict subscription limits. Use the import above when you have real usage.
Daily app dev: debugging, refactoring, multi-step coding tasks across a few sessions per day. ~30 turns per session.
- Cacheable input (system + tools) $0
- Variable input (user + tool results) $0
- Output (incl. tool calls) $0
- Subtotal × retry $0
- Without caching $0
- With caching (current setting) $0
- Caching saves $0
Claude Code reuses system prompts and tool definitions, so prompt caching applies aggressively.
▸ Edit assumptions 10 inputs
This assumes stable system prompts and tool definitions are cacheable across turns. Real cache hit rates vary by session shape and Anthropic's current caching behavior.
This estimate is the API-equivalent value of your usage at Anthropic's published per-token rates. Subscription plans (Pro / Max) are rate-limited, not dollar-billed, so your subscription cost is the plan price plus any optional paid usage credits. Real overage behavior, plan limits, and pricing can change, check claude.com/pricing.
What the plan comparison includes
The baseline uses US web monthly list prices before tax. Pro is $20/month, Max 5x is $100/month, and Max 20x is $200/month. Annual Pro billing is $200 up front. Plan prices checked 2026-10-27 against Claude pricing.
Limits apply across Claude and Claude Code. You can optionally pay for usage beyond your plan limits. Max also includes separate API credits and other benefits that this token-only comparison does not value. Review the current plan and limit rules before changing your subscription.
A capped wrap-up allowance may let an in-progress Claude Code response finish after a five-hour session limit. It still counts toward your weekly limit. On Pro it can apply once per weekly period; on Max it can apply at each session limit within the weekly limit. Rollout varies by account. Usage rules checked 2026-10-27.
The API comparison uses standard-speed global token rates. Fast mode, US-only inference premiums and server-side tool fees are excluded from the imported token estimate.
The report only sees the local records available to ccusage on the computer where you run it. Deleted logs, another computer, or other Claude app activity can make it incomplete. A past calendar month is treated as complete for comparison, but this tool cannot verify that your logs cover the entire month.
Claude model prices checked 2026-10-27 against Anthropic's API pricing. Each model's own verification date is available in the pricing dataset.
Questions about the numbers
Is this my Claude bill?
No. This reprices the token counts recorded on this computer at published API list rates. It does not reconstruct your payments or include tax, paid usage credits, Claude app usage, or the value of other plan benefits.
Can monthly token counts tell me which plan has enough capacity?
No. Claude Code shares session and weekly limits with the Claude apps. Monthly totals do not show usage bursts or other app activity. Check /usage in Claude Code and your actual interruptions before changing tiers.
Why choose a cache-write lifetime?
Aggregate ccusage reports give one cache-creation total without the original request-level lifetime. Five-minute API writes cost 1.25 times normal input; one-hour writes cost twice normal input. Try both as API scenarios. Neither setting reconstructs your subscription bill.
What happens when a model is unknown?
We retain its token counts and any ccusage cost estimate, label the uncertainty, and withhold the monthly verdict and receipt. Unknown prices are not treated as free usage. Models with request-dependent price tiers also need request-level data for a reliable total.
What does the history card assume?
It reprices every recorded month at the current table and compares each month with the same selected plan price. It is a hypothetical comparison, not your billing history. Partial or uncertain months cannot be exported as a verified history receipt.
Does my report leave my browser?
No. Parsing, calculations and receipt rendering happen in your browser. We count anonymous actions such as a successful import, without sending the report, model names, token counts, costs, or a visitor ID.
Why might this differ from ccusage?
Different price tables, stored costs, cache-write assumptions, request-level tiers, and missing models can change the total. This tool recomputes known models from token counts and makes uncertainty visible.
Other tools: agent cost simulator · caching ROI · batch comparison.