HELM Arabic - Arabic Mmlu - Arabic Language (High School): leaderboard

HELM Arabic (release v2.2.0) ArabicMMLU Arabic Language (High School): native Arabic multiple-choice questions from ArabicMMLU, which draws on school, university and professional exams across Arabic-speaking countries; 390 questions, exact match. Percent correct, 0-100, higher is better.

Metric: Exact Match (%). Source: crfm.stanford.edu. Saturation forecast: Around March 2027. 41 models tracked.

Top models

#ModelScoreOverall rank
1Claude Opus 4.774.1#51
2Gemini 2.5 Flash (Non-reasoning)69.49#255 (Gemini 2.5 Flash)
3Claude Sonnet 4.668.21#94
4Gemma 4 31B (IT)64.1#178
5GPT-5.4 (2026-03-05)60.77#63
6Gemini 2.5 Flash Lite (Non-reasoning)57.95#432 (Gemini 2.5 Flash Lite)
7GPT-4.1 (2025-04-14)56.41#257
8Mistral Large 356.15#410
9Claude Haiku 4.5 (20251001)55.38#264
10GPT-5.1 (2025-11-13)55.38#107
11Qwen 3.5 397B A17B53.85#154
12Llama 4 Maverick Instruct FP852.82#424
13Qwen 3 Next 80B A3B Instruct52.56#381
14GPT-5.4 Mini (2026-03-17)52.56#180
15ALLaM-7B-Instruct-preview49.49#724

No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.

Interactive version: theaggregate.ai/benchmark?slug=helm-arabic-arabic-mmlu-arabic-language-high-school · How It Works · Data refreshed daily, snapshot 2026-11-05.