HELM Arabic - Mbzuai Human Translated Arabic Mmlu - High School Mathematics: leaderboard

Metric: Exact Match (%). Source: crfm.stanford.edu. 41 models tracked.

Top models

#ModelScore
1Claude Opus 4.773.33
2Gemini 2.5 Flash (Thinking)62.59
3Claude Haiku 4.5 (20251001)61.85
4Mistral Large 361.11
5Qwen 3 Next 80B A3B Instruct60.37
6Gemma 4 31B (IT)59.26
7Llama 4 Maverick Instruct FP857.04
8Qwen 3.5 397B A17B54.81
9GPT-5.4 (2026-03-05)54.81
10DeepSeek V3.154.07
11Gemini 2.5 Flash Lite (Thinking)52.96
12Qwen 2.5 72B Instruct52.59
13Claude Sonnet 4.652.22
14GPT-4.1 Mini51.11
15GPT-4.1 (2025-04-14)48.15

Interactive version: theaggregate.ai/benchmark?slug=helm-arabic-mbzuai-human-translated-arabic-mmlu-high-school-mathematics · How It Works · Data refreshed daily, snapshot 2026-09-19.