HELM Arabic - Mbzuai Human Translated Arabic Mmlu - Machine Learning: leaderboard
Metric: Exact Match (%). Source: crfm.stanford.edu. 41 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.7 | 91.96 |
| 2 | GPT-5.4 (2026-03-05) | 77.68 |
| 3 | Claude Sonnet 4.6 | 72.32 |
| 4 | Gemma 4 31B (IT) | 72.32 |
| 5 | Claude Haiku 4.5 (20251001) | 72.32 |
| 6 | Qwen 3.5 397B A17B | 66.07 |
| 7 | Gemini 2.5 Flash (Thinking) | 65.18 |
| 8 | GPT-4.1 (2025-04-14) | 63.39 |
| 9 | GPT-4.1 Mini | 60.71 |
| 10 | Qwen 2.5 72B Instruct | 59.82 |
| 11 | Qwen 3 Next 80B A3B Instruct | 58.93 |
| 12 | Mistral Large 2 (Nov) Instruct (2411) | 55.36 |
| 13 | Mistral Large 3 | 54.46 |
| 14 | Llama 4 Maverick Instruct FP8 | 53.57 |
| 15 | Llama 4 Scout Instruct | 51.79 |
Interactive version: theaggregate.ai/benchmark?slug=helm-arabic-mbzuai-human-translated-arabic-mmlu-machine-learning · How It Works · Data refreshed daily, snapshot 2026-09-19.