Capability · Cost
Every AI review,
itemized.
Per-critic cost. Per-PR rollup. Per-vendor token transparency. Honest dashes when a vendor SDK can't report. Receipts your finance team would recognize, attached to every gate decision.
| Critic | Tokens (in / out / cached) | Cost |
|---|---|---|
| ✓ Cursor composer-2.5 | 12,440 / 1,820 / 8,200 | $0.078 |
| ✕ Codex gpt-5.5 | 9,180 / 2,440 / 5,600 | $0.043 |
| ✓ Gemini gemini-3.5-flash | 11,200 / 980 / — | — |
| ✓ Grok grok-4.3 | 10,650 / 1,540 / 6,100 | $0.063 |
| PR total | $0.184 3 of 4 priced | |
Why itemized
Opaque invoices end here.
Most AI dev tools obscure cost behind a flat fee or a "you'll figure it out from your invoice" posture. Dark Factory itemizes the bill.
Per-critic breakdown
See which models drive cost on which kinds of work. Cursor reviewed your PR for
$0.078, Codex for $0.043, Gemini for $0.012,
Grok for $0.025.
Honest dashes
When a critic can't report cost (vendor SDK limitation), the cell renders
—, not a fake zero. The PR total says
$0.184 (3 of 4 priced) — false precision avoided.
Pricing-version audit
Every priced row stamps the pricing-table version and markup-policy version. When rates change, historical PRs stay at their original priced version — no silent re-pricing of history.
Get Started
See what your PR reviews cost.
Install Dark Factory and the dashboard surfaces per-PR cost the moment your first verdict lands.