math benchmark
MMLU-ProX
Extended version of MMLU-Pro providing additional challenging multiple-choice questions for evaluating language models across diverse academic and professional domains. Built on the foundation of the Massive Multitask Language Understanding benchmark framework.
Updated Aug 7, 2026
Published models32
Registry coverage32
MetricScore
EvidenceB
MMLU-ProX leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
Score distribution
Top published rows on the benchmark's original scale.
MMLU-ProX highlights
The top published results on this benchmark's own scale.
What is MMLU-ProX?
Definition and scoring fields from the benchmark registry.
Extended version of MMLU-Pro providing additional challenging multiple-choice questions for evaluating language models across diverse academic and professional domains. Built on the foundation of the Massive Multitask Language Understanding benchmark framework.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- MMLU-ProX
- Modality
- text
- Primary category
- math
- Score direction
- higher
- LLMBoard eligible
- Yes
- Evaluation key
- mmlu-prox|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about MMLU-ProX.
Which model scores highest on MMLU-ProX?
Qwen3.7 Max is currently ranked first with 87.0%.
What does MMLU-ProX measure?
Extended version of MMLU-Pro providing additional challenging multiple-choice questions for evaluating language models across diverse academic and professional domains. Built on the foundation of the Massive Multitask Language Understanding benchmark framework.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
32 unique published model results are currently shown.
Does this benchmark affect the overall score?
This benchmark is marked as eligible for the current LLMBoard capability methodology.