HELM Arabic - Aratrust - Offensive: leaderboard
Metric: Exact Match (%). Source: crfm.stanford.edu. 41 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Llama 4 Maverick Instruct FP8 | 95.65 |
| 2 | jais-family-30B-16k-chat | 95.65 |
| 3 | Mistral Large 3 | 94.2 |
| 4 | Gemini 2.5 Flash Lite (Thinking) | 94.2 |
| 5 | Gemini 2.5 Flash (Thinking) | 92.75 |
| 6 | jais-adapted-7B-chat | 92.75 |
| 7 | Claude Haiku 4.5 (20251001) | 91.3 |
| 8 | Llama 3.3 70B Instruct | 88.41 |
| 9 | Gemma 4 31B (IT) | 88.41 |
| 10 | DeepSeek V3.1 | 88.41 |
| 11 | GPT-4.1 Nano | 86.96 |
| 12 | Qwen 3.5 397B A17B | 86.96 |
| 13 | Mistral Large 2 (Nov) Instruct (2411) | 86.96 |
| 14 | AceGPT-v2-32B-Chat | 86.96 |
| 15 | Qwen 2.5 72B Instruct | 85.51 |
Interactive version: theaggregate.ai/benchmark?slug=helm-arabic-aratrust-offensive · How It Works · Data refreshed daily, snapshot 2026-09-19.