The AI arena is free today

Open Superagent

Multilingual MGSM (CoT)

Paper

Progress Over Time

Interactive timeline showing model performance evolution on Multilingual MGSM (CoT)

State-of-the-art frontier
Open
Proprietary

Multilingual MGSM (CoT) Leaderboard

3 models
ContextCostLicense
1405B
270B
38B
Notice missing or incorrect data?
About this benchmark

What is Multilingual MGSM (CoT)?

Multilingual Grade School Math (MGSM) benchmark evaluates language models' chain-of-thought reasoning abilities across ten typologically diverse languages. Contains 250 grade-school math problems manually translated from GSM8K dataset into languages including Bengali and Swahili.

Multilingual MGSM (CoT) is a text benchmark evaluating models on math and reasoning tasks. LLM Stats tracks 3 models on this benchmark, scored on a 0–1 scale. The current average is 0.8, with the leader at 0.9.

Compare leaders on the best AI for math and best AI for reasoning leaderboards.

Current leaders

Llama 3.1 405B Instruct from Meta currently leads the Multilingual MGSM (CoT) leaderboard with a score of 0.916 across 3 evaluated AI models.

Source paper

Title
Language Models are Multilingual Chain-of-Thought Reasoners
Authors
Freda Shi, Mirac Suzgun, Markus Freitag, Xuezhi Wang, and 8 others
Published
Abstract

We evaluate the reasoning abilities of large language models in multilingual settings. We introduce the Multilingual Grade School Math (MGSM) benchmark, by manually translating 250 grade-school math problems from the GSM8K dataset (Cobbe et al., 2021) into ten typologically diverse languages. We find that the ability to solve MGSM problems via chain-of-thought prompting emerges with increasing model scale, and that models have strikingly strong multilingual reasoning abilities, even in underrepresented languages such as Bengali and Swahili. Finally, we show that the multilingual reasoning abilities of language models extend to other tasks such as commonsense reasoning and word-in-context semantic judgment. The MGSM benchmark is publicly available at https://github.com/google-research/url-nlp.

FAQ

Common questions about the Multilingual MGSM (CoT) benchmark and leaderboard.

What is the Multilingual MGSM (CoT) benchmark?

Multilingual Grade School Math (MGSM) benchmark evaluates language models' chain-of-thought reasoning abilities across ten typologically diverse languages. Contains 250 grade-school math problems manually translated from GSM8K dataset into languages including Bengali and Swahili.

What is the Multilingual MGSM (CoT) leaderboard?

The Multilingual MGSM (CoT) leaderboard ranks 3 AI models based on their performance on this benchmark. Currently, Llama 3.1 405B Instruct by Meta leads with a score of 0.916. The average score across all models is 0.825.

What is the highest Multilingual MGSM (CoT) score?

The highest Multilingual MGSM (CoT) score is 0.916, achieved by Llama 3.1 405B Instruct from Meta.

How many models are evaluated on Multilingual MGSM (CoT)?

3 models have been evaluated on the Multilingual MGSM (CoT) benchmark, with 0 verified results and 3 self-reported results.

Where can I find the Multilingual MGSM (CoT) paper?

The Multilingual MGSM (CoT) paper is available at https://arxiv.org/abs/2210.03057. The paper details the methodology, dataset construction, and evaluation criteria.

What categories does Multilingual MGSM (CoT) cover?

Multilingual MGSM (CoT) is categorized under math and reasoning. The benchmark evaluates text models with multilingual support.

What is the best open-source model on Multilingual MGSM (CoT)?

Llama 3.1 405B Instruct by Meta is the top-ranked open-source model on Multilingual MGSM (CoT), with a score of 0.916 (rank #1).

How recent are the Multilingual MGSM (CoT) leaderboard results?

The Multilingual MGSM (CoT) leaderboard was last updated in August 2026 and currently includes 3 evaluated models.