Arabic IFEval — leaderboard
Arabic instruction-following evaluation testing prompt-level and instruction-level accuracy in Arabic, adapted from the IFEval benchmark.
Metric: Arabic Accuracy (%). Source: huggingface.co. Status: saturation imminent. 40 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude 3.5 Sonnet | 75.9 |
| 2 | GPT-4o (2024-08-06) | 74.4 |
| 3 | GPT-4o Mini (2024-07-18) | 71.4 |
| 4 | Claude 3.5 Haiku | 70.9 |
| 5 | Qwen 2.5 72B Instruct | 70.7 |
| 6 | Qwen 2.5 32B Instruct | 63.7 |
| 7 | Gemma 2 27B (IT) | 63.5 |
| 8 | aya-expanse-32B | 61.3 |
| 9 | c4ai-command-r7B-12-2024 | 60.8 |
| 10 | Llama 3.3 70B Instruct | 60.7 |
| 11 | Claude 3 Haiku | 59 |
| 12 | Gemma 2 9B (IT) | 56 |
| 13 | Llama 3.1 405B Instruct | 53.8 |
| 14 | aya-expanse-8B | 53.5 |
| 15 | Qwen 2.5 7B Instruct | 52.8 |
Interactive version: theaggregate.ai/benchmark?slug=arabic-ifeval · How the rankings work · Data refreshed daily, snapshot 2026-07-22.