Knowledge · Benchmark profile
A filtered hard subset of KMMLU containing ~5,000 questions that most models get wrong.
Data verified 21 Jul 2026 · Methodology 1.6.0
This benchmark is registered, but the current snapshot has no qualifying results for the model cohort.
Not evaluated for the selected model cohortAbout KMMLU-Hard
A filtered hard subset of KMMLU containing ~5,000 questions that most models get wrong. Results stay tied to the exact model variant and evaluation system. Multiple systems for the same model use the best published score on this page; overall Lumina scoring uses the median of ranking-eligible rows.
Open benchmark source ↗FAQ
A filtered hard subset of KMMLU containing ~5,000 questions that most models get wrong.
No models have qualifying results in the current snapshot.
0 models in the LuminaBench cohort have a qualifying score on this benchmark.
Related
2026 · 10 results · Reference
Artificial Analysis MMLU-Pro2026 · 4 results · Reference
Massive Multitask Language Understanding2020 · 8 results · Reference
Graduate-Level Google-Proof Q&A2023 · 70 results · Reference
SuperGPQA: Scaling LLM Evaluation Across 285 Graduate Disciplines2025 · 19 results · Reference
Massive Multitask Language Understanding Professional2024 · 42 results · Reference
