all the models — AI benchmark observatory
← Benchmarks

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.

id mgsm · max 1 · 31 models reported

#ModelScore
1Llama 4 Maverick
Meta · open
0.92
2o3-mini
OpenAI
0.92
3Claude 3.5 Sonnet
Anthropic
0.92
4Claude 3.5 Sonnet
Anthropic
0.92
5Llama 3.3 70B Instruct
Meta · open
0.91
6o1-preview
OpenAI
0.91
7Claude 3 Opus
Anthropic
0.91
8Llama 4 Scout
Meta · open
0.91
9GPT-4o
OpenAI
0.91
10o1
OpenAI
0.89