HELM Arabic - Mbzuai Human Translated Arabic Mmlu: leaderboard
The "Mbzuai Human Translated Arabic Mmlu" subtask of HELM Arabic. Stanford HELM's Arabic project: Modern Standard Arabic knowledge, reasoning and understanding scenarios (AlGhafa, Arabic MMLU, exams and native-authored sets) run through the HELM harness with exact-match style metrics.
Metric: EM. Source: crfm.stanford.edu. Saturation forecast: Around April 2027. 41 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.7 | 87.96 | #50 |
| 2 | GPT-5.4 (2026-03-05) | 83.18 | #58 |
| 3 | Claude Sonnet 4.6 | 82.97 | #92 |
| 4 | Gemini 2.5 Flash (Non-reasoning) | 80.56 | #242 (Gemini 2.5 Flash) |
| 5 | Gemma 4 31B (IT) | 80.37 | #178 |
| 6 | Qwen 3.5 397B A17B | 80.28 | #147 |
| 7 | Claude Haiku 4.5 (20251001) | 79.05 | #258 |
| 8 | GPT-5.1 (2025-11-13) | 78.41 | #105 |
| 9 | GPT-4.1 (2025-04-14) | 78.2 | #250 |
| 10 | Mistral Large 3 | 77.31 | #404 |
| 11 | Llama 4 Maverick Instruct FP8 | 76.43 | #412 |
| 12 | Qwen 3 Next 80B A3B Instruct | 75.07 | #370 |
| 13 | DeepSeek V3.1 | 74.21 | #268 |
| 14 | GPT-5.4 Mini (2026-03-17) | 74.18 | #171 |
| 15 | Qwen 2.5 72B Instruct | 72.94 | #458 |
No result here: #4 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=helm-arabic-mbzuai-human-translated-arabic-mmlu · How It Works · Data refreshed daily, snapshot 2026-10-23.