HELM Arabic - Alghafa - MCQ Exams Test AR: leaderboard
HELM Arabic (release v2.2.0) AlGhafa MCQ Exams Test AR task: four-option Arabic school exam questions, from the native AlGhafa Arabic multiple-choice benchmark; 557 items, exact match. Percent correct, 0-100, higher is better.
Metric: Exact Match (%). Source: crfm.stanford.edu. Saturation forecast: Around December 2027. 41 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 3.3 70B Instruct | 64.99 | #541 |
| 2 | Claude Opus 4.7 | 62.66 | #51 |
| 3 | GPT-4.1 (2025-04-14) | 61.58 | #257 |
| 4 | GPT-5.1 (2025-11-13) | 61.58 | #107 |
| 5 | Claude Sonnet 4.6 | 61.4 | #94 |
| 6 | GPT-5.4 (2026-03-05) | 61.4 | #63 |
| 7 | Llama 4 Maverick Instruct FP8 | 60.86 | #424 |
| 8 | Gemini 2.5 Flash (Non-reasoning) | 60.32 | #255 (Gemini 2.5 Flash) |
| 9 | Llama 4 Scout Instruct | 59.96 | #547 |
| 10 | Qwen 3.5 397B A17B | 59.78 | #154 |
| 11 | Mistral Large 3 | 59.78 | #410 |
| 12 | Gemini 2.5 Flash Lite (Non-reasoning) | 59.61 | #432 (Gemini 2.5 Flash Lite) |
| 13 | Qwen 2.5 72B Instruct | 59.43 | #469 |
| 14 | Gemma 4 31B (IT) | 59.25 | #178 |
| 15 | Claude Haiku 4.5 (20251001) | 59.07 | #264 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=helm-arabic-alghafa-mcq-exams-test-ar · How It Works · Data refreshed daily, snapshot 2026-11-05.