Evals for Every Language - Arabic (ar): leaderboard
Average score across evaluation tasks in Arabic (ar), from Evals for Every Language.
Metric: Average Score (%). Source: huggingface.co. Saturation forecast: Around June 2027. 67 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.8 | 76.28 | #37 |
| 2 | Claude Opus 4.7 | 75.1 | #49 |
| 3 | Gemini 3.1 Pro (Preview) | 73.25 | #58 |
| 4 | Claude Sonnet 4.5 | 72.84 | #142 |
| 5 | Gemini 2.5 Pro | 72.58 | #152 |
| 6 | DeepSeek V3.1 | 72.31 | #264 |
| 7 | Grok 4.20 | 71.34 | #175 |
| 8 | Gemini 3.1 Flash Lite | 71.21 | #191 |
| 9 | GPT-5 | 71.18 | #98 |
| 10 | Claude Sonnet 4 | 70.62 | #199 |
| 11 | Claude Haiku 4.5 | 70.41 | #280 |
| 12 | GPT-5.1 | 69.21 | #135 |
| 13 | GPT-5.3 | 68.44 | #103 |
| 14 | Claude Sonnet 4.6 | 68.34 | #90 |
| 15 | jamba-large-1.7 | 68.3 | #525 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=evals-for-every-language-language-ar · How It Works · Data refreshed daily, snapshot 2026-10-20.