reasoning benchmark
CMMLU
CMMLU (Chinese Massive Multitask Language Understanding) is a comprehensive Chinese benchmark that evaluates the knowledge and reasoning capabilities of large language models across 67 different subject topics. The benchmark covers natural sciences, social sciences, engineering, and humanities with multiple-choice questions ranging from basic to advanced professional levels.
Updated Aug 7, 2026
Published models6
Registry coverage6
MetricScore
EvidenceB
CMMLU leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | XI | 90.2% | 100.0% | 6 | C | |
| 2 | AC | 90.1% | 80.0% | 6 | C | |
| 3 | ME | 84.3% | 60.0% | 6 | C | |
| 4 | ME | 82.5% | 40.0% | 6 | C | |
| 5 | OP | 81.5% | 20.0% | 6 | C | |
| 6 | BA | 39.8% | 0.0% | 6 | C |
Score distribution
Top published rows on the benchmark's original scale.
CMMLU highlights
The top published results on this benchmark's own scale.
What is CMMLU?
Definition and scoring fields from the benchmark registry.
CMMLU (Chinese Massive Multitask Language Understanding) is a comprehensive Chinese benchmark that evaluates the knowledge and reasoning capabilities of large language models across 67 different subject topics. The benchmark covers natural sciences, social sciences, engineering, and humanities with multiple-choice questions ranging from basic to advanced professional levels.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- CMMLU
- Modality
- text
- Primary category
- reasoning
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- cmmlu|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about CMMLU.
Which model scores highest on CMMLU?
MiMo-V2.5-Pro is currently ranked first with 90.2%.
What does CMMLU measure?
CMMLU (Chinese Massive Multitask Language Understanding) is a comprehensive Chinese benchmark that evaluates the knowledge and reasoning capabilities of large language models across 67 different subject topics. The benchmark covers natural sciences, social sciences, engineering, and humanities with multiple-choice questions ranging from basic to advanced professional levels.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
6 unique published model results are currently shown.
Does this benchmark affect the overall score?
This benchmark is preserved as source-native evidence but is not eligible for the current overall score.