Enter your monthly PR count, developer count, and average changed files per PR to get a side-by-side monthly cost estimate, based on Orange Codens' measured per-token pricing and CodeRabbit's published pricing and Fair Usage policy.
This simulator produces estimates based on each vendor's official published pricing and does not guarantee your actual bill. CodeRabbit's Fair Usage policy, per its own documentation, progressively throttles review speed rather than publishing a hard monthly cap. We model it here using the documented example — roughly 60 reviews per 7 days per developer (≈257/month) — as a monthly allowance. Your actual invoice from either vendor depends on your specific contract and usage. Always check each vendor's official pricing page for the latest numbers.
Used only for CodeRabbit's usage-based add-on ($0.25 / reviewed file).
Default ¥150/$1 is an approximation as of July 2026. Adjust to your actual rate.
Fixed monthly price of the matching plan (as long as its included credits cover your usage)
Base fee + usage-based add-on (modeled estimate — your actual invoice may differ)
The calculation logic and constants live in pricing-simulator.js (built so the numbers can be updated in one place).
The per-review cost for "Small", "Medium", and "Large" PR is taken directly from the BCP (billing control plane) billing ledger (operation_type='orange.review', n=1,273, 2026-07-08 to 2026-07-27) — actual charged credit percentiles, not a token-based estimate. Small ≈16.4 credits (p25, about JPY 0.16); Medium ≈78.8 credits (p50/median, about JPY 0.79); Large ≈257.6 credits (p90, top 10% by diff size, about JPY 2.58). All three tiers are directly measured from the same ledger — none are extrapolated. We don't back-calculate from token counts via the rate card (input 0.0225 / output 0.1125 credits/token, qwen multiplier 0.1) — real usage includes some reviews on models other than qwen, so the actual average differs from that theoretical figure. We then present the cheapest plan whose included credits cover the estimated usage (Hobby JPY 3,000 = 300K credits / Pro JPY 10,000 = 1M credits / Business JPY 50,000 = 5M credits).
Being upfront about the tail: the same ledger's upper percentiles run p95 ≈438.4 credits (about JPY 4.38), p99 ≈4,629.2 credits (about JPY 46.29), and a measured max of ≈15,176.6 credits (about JPY 151.77). Orange's credit-based pricing has no per-review cost cap — the flip side is that a small share of very large PRs cost substantially more than the median. We'd rather disclose that than hide it.
We take the Pro plan base fee ($24/dev/month annual, or $30/dev/month monthly) and treat the documented Fair Usage example (60 reviews / 7 days / developer ≈ 257/month) as an org-wide monthly allowance (developers × 257), adding a usage-based add-on ($0.25 × files reviewed) for anything past that. Fair Usage is officially a progressive throttle, not a published hard monthly cap, so this calculation is our own modeling. Your actual invoice depends on how CodeRabbit applies its policy.
The default ¥150/$1 is an approximation as of July 2026. Adjust it freely in the form above.