math benchmark
MMLU (CoT)
Chain-of-Thought variant of the Massive Multitask Language Understanding benchmark, evaluating language models across 57 tasks including elementary mathematics, US history, computer science, law, and other professional and academic subjects. This version uses chain-of-thought prompting to elicit step-by-step reasoning.
Updated Aug 7, 2026
Published models3
Registry coverage3
MetricScore
EvidenceB
MMLU (CoT) leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | ME | 88.6% | 100.0% | 3 | C | |
| 2 | ME | 86.0% | 50.0% | 3 | C | |
| 3 | ME | 73.0% | 0.0% | 3 | C |
Score distribution
Top published rows on the benchmark's original scale.
MMLU (CoT) highlights
The top published results on this benchmark's own scale.
What is MMLU (CoT)?
Definition and scoring fields from the benchmark registry.
Chain-of-Thought variant of the Massive Multitask Language Understanding benchmark, evaluating language models across 57 tasks including elementary mathematics, US history, computer science, law, and other professional and academic subjects. This version uses chain-of-thought prompting to elicit step-by-step reasoning.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- MMLU (CoT)
- Modality
- text
- Primary category
- math
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- mmlu-(cot)|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about MMLU (CoT).
Which model scores highest on MMLU (CoT)?
Llama 3.1 405B Instruct is currently ranked first with 88.6%.
What does MMLU (CoT) measure?
Chain-of-Thought variant of the Massive Multitask Language Understanding benchmark, evaluating language models across 57 tasks including elementary mathematics, US history, computer science, law, and other professional and academic subjects. This version uses chain-of-thought prompting to elicit step-by-step reasoning.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
3 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.