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

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

Top models

#ModelScore
1Claude Opus 4.794.44
2Claude Sonnet 4.692.93
3Qwen 3.5 397B A17B89.9
4GPT-5.4 (2026-03-05)89.9
5GPT-4.1 (2025-04-14)89.39
6Claude Haiku 4.5 (20251001)88.38
7Mistral Large 388.38
8Gemini 2.5 Flash (Thinking)88.38
9Llama 4 Maverick Instruct FP886.87
10Gemma 4 31B (IT)85.35
11Qwen 3.5 9B84.34
12Qwen 3 Next 80B A3B Instruct84.34
13AceGPT-v2-70B-Chat84.34
14Llama 3.3 70B Instruct83.84
15DeepSeek V3.183.84

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