HELM Arabic Enterprise - Arabic Content Generation: leaderboard
Metric: Score. Source: crfm.stanford.edu. 35 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemma 4 31B (IT) | 87.5 |
| 2 | GPT-5.4 (2026-03-05) | 86.41 |
| 3 | DeepSeek V4 Pro | 85.1 |
| 4 | Gemini 2.5 Flash (Thinking) | 80.14 |
| 5 | Claude Opus 4.7 | 79.02 |
| 6 | Llama 4 Maverick Instruct | 78.34 |
| 7 | Qwen3 235B A22B Instruct 2507 FP8 | 77.52 |
| 8 | DeepSeek V3.1 | 77.36 |
| 9 | Gemini 2.5 Flash Lite (Thinking) | 75.71 |
| 10 | Claude Haiku 4.5 (20251001) | 75.68 |
| 11 | Qwen 3.5 9B | 73.76 |
| 12 | Llama 4 Scout Instruct | 72.64 |
| 13 | Claude Sonnet 4.6 | 72.37 |
| 14 | Llama 3.3 70B Instruct | 68.81 |
| 15 | AceGPT-v2-70B-Chat | 68.81 |
Interactive version: theaggregate.ai/benchmark?slug=helm-arabic-enterprise-arabic-content-generation · How It Works · Data refreshed daily, snapshot 2026-09-08.