HELM Arabic - Alrage: leaderboard
The "Alrage" 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: Score. Source: crfm.stanford.edu. Saturation forecast: Around May 2027. 41 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.4 (2026-03-05) | 92.33 | #58 |
| 2 | Gemma 4 31B (IT) | 91.63 | #179 |
| 3 | Qwen 3.5 397B A17B | 91.51 | #148 |
| 4 | GPT-5.4 Mini (2026-03-17) | 91.04 | #173 |
| 5 | GPT-5.1 (2025-11-13) | 90.88 | #106 |
| 6 | Gemini 2.5 Flash (Non-reasoning) | 89.55 | #247 (Gemini 2.5 Flash) |
| 7 | Qwen 3.5 9B | 89.51 | #388 |
| 8 | Qwen 3 Next 80B A3B Instruct | 89 | #373 |
| 9 | GPT-4.1 (2025-04-14) | 88.96 | #252 |
| 10 | Gemini 2.5 Flash Lite (Non-reasoning) | 88.18 | #426 (Gemini 2.5 Flash Lite) |
| 11 | GPT-5.4 Nano (2026-03-17) | 87.66 | #270 |
| 12 | DeepSeek V3.1 | 87.62 | #268 |
| 13 | GPT-4.1 Mini | 87.4 | #358 |
| 14 | Mistral Large 3 | 85.46 | #409 |
| 15 | ALLaM-7B-Instruct-preview | 85.21 | #725 |
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-alrage · How It Works · Data refreshed daily, snapshot 2026-10-24.