HELM Arabic - Mbzuai Human Translated Arabic Mmlu - High School Government And Politics: leaderboard

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

Top models

#ModelScore
1Claude Sonnet 4.696.89
2Claude Opus 4.796.37
3Gemini 2.5 Flash (Thinking)96.37
4Qwen 3.5 397B A17B95.34
5GPT-5.4 (2026-03-05)95.34
6Claude Haiku 4.5 (20251001)92.75
7GPT-4.1 (2025-04-14)91.19
8AceGPT-v2-70B-Chat90.67
9Gemma 4 31B (IT)90.16
10Mistral Large 390.16
11DeepSeek V3.189.12
12Qwen 2.5 72B Instruct88.6
13Llama 4 Maverick Instruct FP887.56
14Llama 3.3 70B Instruct86.53
15Qwen 3 Next 80B A3B Instruct86.53

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