HELM Arabic - Arabic Mmlu - Arabic Language (General): leaderboard
Metric: Exact Match (%). Source: crfm.stanford.edu. 41 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.7 | 93.14 |
| 2 | Claude Sonnet 4.6 | 91.67 |
| 3 | Gemini 2.5 Flash (Thinking) | 90.85 |
| 4 | GPT-5.4 (2026-03-05) | 89.54 |
| 5 | Gemma 4 31B (IT) | 89.22 |
| 6 | GPT-4.1 (2025-04-14) | 89.05 |
| 7 | Gemini 2.5 Flash Lite (Thinking) | 88.4 |
| 8 | Mistral Large 3 | 88.24 |
| 9 | Llama 4 Maverick Instruct FP8 | 88.07 |
| 10 | Claude Haiku 4.5 (20251001) | 87.42 |
| 11 | Qwen 3.5 397B A17B | 87.09 |
| 12 | Llama 4 Scout Instruct | 85.13 |
| 13 | Qwen 3 Next 80B A3B Instruct | 84.48 |
| 14 | GPT-4.1 Mini | 83.82 |
| 15 | Qwen 2.5 72B Instruct | 83.33 |
Interactive version: theaggregate.ai/benchmark?slug=helm-arabic-arabic-mmlu-arabic-language-general · How It Works · Data refreshed daily, snapshot 2026-09-19.