Skip to content

Leaderboard

5 results · score desc, then earliest result

Benchmark leaderboard
#ScoreTaskAdapterModelSpecSecMaintPass@kReliabilityFlagsCreated
199.5μ 86.3 · σ 18.7taskboard
Fake fixture (taskboard/compliant)Sample
no model100.0100.096.767%

3 runs · 67% pass

—15 Nov 2026, 15:38
299.5μ 86.3 · σ 18.7taskboard
Fake fixture (taskboard/compliant)Sample
no model100.0100.096.767%

3 runs · 67% pass

—18 Nov 2026, 06:08
399.3μ 74.6 · σ 24.7linkstash
Fake fixture (linkstash/compliant)Sample
no model100.0100.095.050%

2 runs · 50% pass

—15 Nov 2026, 15:38
459.9μ 86.3 · σ 18.7taskboard
Fake fixture (taskboard/regressed)Sample
no model52.4100.090.067%

3 runs · 67% pass

Capped15 Nov 2026, 15:38
549.9μ 74.6 · σ 24.7linkstash
Fake fixture (linkstash/regressed)Sample
no model50.033.390.050%

2 runs · 50% pass

Capped15 Nov 2026, 15:38

ids: res_1ab24b…662a · res_6d753e…2b41 · res_bb227a…20ef · res_0e0868…71e1 · res_2ac581…2981

A leaderboard-valid row records task, ruleset, scoring-config, and adapter versions plus a repository fingerprint. Rows missing any of them are rejected by the store schema.