math benchmark
MGSM
MGSM (Multilingual Grade School Math) is a benchmark of grade-school math problems. Contains 250 grade-school math problems manually translated from the GSM8K dataset into ten typologically diverse languages: Spanish, French, German, Russian, Chinese, Japanese, Thai, Swahili, Bengali, and Telugu. Evaluates multilingual mathematical reasoning capabilities.
Updated Aug 7, 2026
Published models31
Registry coverage31
MetricScore
EvidenceB
MGSM 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.
MGSM highlights
The top published results on this benchmark's own scale.
What is MGSM?
Definition and scoring fields from the benchmark registry.
MGSM (Multilingual Grade School Math) is a benchmark of grade-school math problems. Contains 250 grade-school math problems manually translated from the GSM8K dataset into ten typologically diverse languages: Spanish, French, German, Russian, Chinese, Japanese, Thai, Swahili, Bengali, and Telugu. Evaluates multilingual mathematical reasoning capabilities.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- MGSM
- Modality
- text
- Primary category
- math
- Score direction
- higher
- LLMBoard eligible
- Yes
- Evaluation key
- mgsm|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about MGSM.
Which model scores highest on MGSM?
Llama 4 Maverick is currently ranked first with 92.3%.
What does MGSM measure?
MGSM (Multilingual Grade School Math) is a benchmark of grade-school math problems. Contains 250 grade-school math problems manually translated from the GSM8K dataset into ten typologically diverse languages: Spanish, French, German, Russian, Chinese, Japanese, Thai, Swahili, Bengali, and Telugu. Evaluates multilingual mathematical reasoning capabilities.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
31 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.