Wikibench

MMMLU

math

Multilingual Massive Multitask Language Understanding dataset released by OpenAI, featuring professionally translated MMLU test questions across 14 languages including Arabic, Bengali, German, Spanish, French, Hindi, Indonesian, Italian, Japanese, Korean, Portuguese, Swahili, Yoruba, and Chinese. Contains approximately 15,908 multiple-choice questions per language covering 57 subjects.

Leaderboard

Showing 20 of 49 results

  1. Claude Mythos Preview
    92.7%
    i
  2. Gemini 3.1 Pro
    92.6%
    i
  3. Gemini 3 Flash
    91.8%
    i
  4. Gemini 3 Pro
    91.8%
    i
  5. Claude Opus 4.7
    91.5%
    i
  6. Claude Opus 4.6
    91.1%
    i
  7. Claude Opus 4.5
    90.8%
    i
  8. Qwen3.7 Max
    90.3%
    i
  9. GPT-5.2
    89.6%
    i
  10. Claude Opus 4.1
    89.5%
    i
  11. Qwen3.6 Plus
    89.5%
    i
  12. Claude Sonnet 4.6
    89.3%
    i
  13. Claude Sonnet 4.5
    89.1%
    i
  14. Gemini 3.1 Flash-Lite
    88.9%
    i
  15. Claude Opus 4
    88.8%
    i
  16. Qwen3.5-397B-A17B
    88.5%
    i
  17. Gemma 4 31B
    88.4%
    i
  18. o1
    87.7%
    i
  19. GPT-4.1
    87.3%
    i
  20. Qwen3 235B A22B
    86.7%
    i