A scorecard you can defend
For buyers without robots: one signed report compares the bodies for you.
One signed report instead of three robots
- I pay
- $8k–15k per report i
- I get
- A signed document I can take to my boss, board or insurer, plus GAP and cross-body variance, which no one else sells
- I never see
- B-tier tasks, or my own share of B-tier footage. Ever, even paid. i
Choose what gets tested
One body = one table row. Cross-body variance needs at least 3.
The private premium is the price of RC's own lottery ticket.i
Results confidential · you own publication · RC keeps anonymized aggregate rows forever · raw data stays at RC; you receive the report
Your price updates as you pick ILLUSTRATIVE
What you receive: the gap nobody else can compute ILLUSTRATIVE
Protocol, objects and scripts all public; anyone can self-test
Never published i
How much the policy was tuned to the public benchmark. Only someone holding both sides can compute it.
Why buyers pay for this page
They aren't buying the percentages; those are available elsewhere. They buy the signature, and the GAP and cross-body variance they can't produce themselves.
Attempt and success, never multiplied
attempt_rate · language layer
Will the model try? Nearly flat across bodies.i
success_rate · body layer
Does it succeed? Varies widely across bodies. Multiplying the two into one "success rate" is what the industry does today.i