Benchmark
MMMLU Benchmark
What is MMMLU?
Multilingual Massive Multitask Language Understanding dataset released by OpenAI, featuring professionally translated MMLU test questions across 14 languages…
- Modality
- text
- Evaluates
- math, reasoning, language, general
- Openness
- open
- Importer
- LLM Stats
- Reported scores
- 49
Source provenance
- Original evidence https://api.zeroeval.com/leaderboard/benchmarks/mmmlu?top_n=500
MMMLU code and dataset
MMMLU results and reported scores
LLM Stats
| Model | Organization | Reported value | Reported | Evidence |
|---|---|---|---|---|
| Claude 3.7 Sonnet | Anthropic | 0.861 | 2025-02-24 | evidence |
| Claude Haiku 4.5 | Anthropic | 0.83 | 2025-10-15 | evidence |
| Claude Mythos Preview | Anthropic | 0.927 | 2026-04-07 | evidence |
| Claude Opus 4 | Anthropic | 0.888 | 2025-05-22 | evidence |
| Claude Opus 4.1 | Anthropic | 0.895 | 2025-08-05 | evidence |
| Claude Opus 4.5 | Anthropic | 0.908 | 2025-11-24 | evidence |
| Claude Opus 4.6 | Anthropic | 0.911 | 2026-02-05 | evidence |
| Claude Opus 4.7 | Anthropic | 0.915 | 2026-04-16 | evidence |
| Claude Sonnet 4 | Anthropic | 0.865 | 2025-05-22 | evidence |
| Claude Sonnet 4.5 | Anthropic | 0.891 | 2025-09-29 | evidence |
| Claude Sonnet 4.6 | Anthropic | 0.893 | 2026-02-17 | evidence |
| DiffusionGemma 26B-A4B | 0.815 | 2026-06-10 | evidence | |
| Gemini 3 Flash | 0.918 | 2025-12-17 | evidence | |
| Gemini 3 Pro | 0.918 | 2025-11-18 | evidence | |
| Gemini 3.1 Flash-Lite | 0.889 | 2026-03-03 | evidence | |
| Gemini 3.1 Pro | 0.926 | 2026-02-19 | evidence | |
| Gemma 4 12B | 0.834 | 2026-05-23 | evidence | |
| Gemma 4 26B-A4B | 0.863 | 2026-04-02 | evidence | |
| Gemma 4 31B | 0.884 | 2026-04-02 | evidence | |
| Gemma 4 E2B | 0.674 | 2026-04-02 | evidence | |
| Gemma 4 E4B | 0.766 | 2026-04-02 | evidence | |
| GPT OSS 120B High | OpenAI | 0.838 | 2025-08-05 | evidence |
| GPT-4.1 | OpenAI | 0.873 | 2025-04-14 | evidence |
| GPT-4.1 mini | OpenAI | 0.785 | 2025-04-14 | evidence |
| GPT-4.1 nano | OpenAI | 0.669 | 2025-04-14 | evidence |
| GPT-4.5 | OpenAI | 0.851 | 2025-02-27 | evidence |
| GPT-4o | OpenAI | 0.814 | 2024-08-06 | evidence |
| GPT-5.2 | OpenAI | 0.896 | 2025-12-11 | evidence |
| K-EXAONE-236B-A23B | LG AI Research | 0.857 | 2025-12-31 | evidence |
| Mistral Large 3 | Mistral | 0.742 | 2025-09-01 | evidence |
| Mistral Large 3 (675B Base) | Mistral | 0.855 | 2025-12-04 | evidence |
| Mistral Large 3 (675B Instruct 2512 Eagle) | Mistral | 0.855 | 2025-12-04 | evidence |
| Mistral Large 3 (675B Instruct 2512 NVFP4) | Mistral | 0.855 | 2025-12-04 | evidence |
| Mistral Large 3 (675B Instruct 2512) | Mistral | 0.855 | 2025-12-04 | evidence |
| o1 | OpenAI | 0.877 | 2024-12-17 | evidence |
| Phi-3.5-mini-instruct | Microsoft | 0.554 | 2024-08-23 | evidence |
| Phi-3.5-MoE-instruct | Microsoft | 0.699 | 2024-08-23 | evidence |
| Qwen3 235B A22B | Qwen | 0.867 | 2025-04-29 | evidence |
| Qwen3.5-0.8B | Qwen | 0.443 | 2026-03-02 | evidence |
| Qwen3.5-122B-A10B | Qwen | 0.867 | 2026-02-24 | evidence |
| Qwen3.5-27B | Qwen | 0.859 | 2026-02-24 | evidence |
| Qwen3.5-2B | Qwen | 0.631 | 2026-03-02 | evidence |
| Qwen3.5-35B-A3B | Qwen | 0.852 | 2026-02-24 | evidence |
| Qwen3.5-397B-A17B | Qwen | 0.885 | 2026-02-16 | evidence |
| Qwen3.5-4B | Qwen | 0.761 | 2026-03-02 | evidence |
| Qwen3.5-9B | Qwen | 0.812 | 2026-03-02 | evidence |
| Qwen3.6 Plus | Qwen | 0.895 | 2026-04-02 | evidence |
| Qwen3.7 Max | Qwen | 0.903 | 2026-05-19 | evidence |
| Qwen3.7-Plus | Qwen | 0.89 | 2026-05-31 | evidence |
Scores are partitioned by the source that reported them and are never merged into a single cross-source ranking, because the sources measure different things and say so.