Benchmark

EnigmaEval

Measures: 1,161 puzzle-hunt puzzles (many multimodal, with grids, images and diagrams, including meta-puzzles) that need hidden connections and multi-step deduction; scored as pass@1 accuracy with 95% CI.

Maintainer
Scale AI (Scale Labs)
Introduced
2025
Status
Active
Saturation
Headroom
Typical provenance
Independent
In scoring set
No

Not scored: Questions were open-sourced in July 2026 (contamination risk), the newest frontier models are not listed, and the Scale page returned 404 on 2026-09-29.

PaperOfficial leaderboard

Every result we hold, whoever ran it. Only independent and third-party rows count towards a score.

No results to show.

EnigmaEval: leaderboard and what it measures · supermarket.dev