Leaderboard

[ESC] MAIN MENU

════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════

TESTTYPE

#	Subject	Type	Test	Effort	Runs	Mean	Spread	Distribution	Latest
1	▸ claude-opus-5	LLM	Booklet A	high	3	87.4% 39.3/45	±1 87–89%	██████████░░	2026-08-10
2	▸ claude-fable-5	LLM	Booklet A	max	4	85% 38.3/45	±2.4 82–89%	██████████░░	2026-08-12
3	▸ claude-sonnet-5	LLM	Booklet A	max	3	67.4% 30.3/45	±1 67–69%	████████░░░░	2026-08-13
4	▸ claude-opus-5	LLM	Booklet A	low	3	66.7% 30/45	±4.8 60–71%	████████░░░░	2026-08-11
5	▸ claude-fable-5	LLM	Booklet A	low	3	60% 27/45	±4.8 56–67%	███████░░░░░	2026-08-10

Each row is the mean of that subject’s sittings at one effort on one test, with the standard deviation and the range across those sittings. Chance on a 6-alternative forced choice is 17% — roughly 8/45. A score near that is indistinguishable from guessing.