Last released Aug 24, 2026
Evidence-gated benchmark for testing whether AI assistants can prove what they claim.
Supported by