HELM Arabic - Alghafa: leaderboard
Metric: EM. Source: crfm.stanford.edu. 41 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.7 | 79.97 |
| 2 | Claude Sonnet 4.6 | 79.72 |
| 3 | GPT-4.1 (2025-04-14) | 79.49 |
| 4 | Gemma 4 31B (IT) | 78.79 |
| 5 | Llama 3.3 70B Instruct | 78.53 |
| 6 | GPT-5.4 (2026-03-05) | 78.44 |
| 7 | Llama 4 Maverick Instruct FP8 | 78.14 |
| 8 | Gemini 2.5 Flash (Thinking) | 78.03 |
| 9 | Gemini 2.5 Flash Lite (Thinking) | 77.63 |
| 10 | Llama 4 Scout Instruct | 77.54 |
| 11 | Claude Haiku 4.5 (20251001) | 77.35 |
| 12 | Qwen3 235B A22B Instruct 2507 FP8 | 77.27 |
| 13 | Qwen 2.5 72B Instruct | 77.18 |
| 14 | Mistral Large 3 | 76.89 |
| 15 | Qwen 3 Next 80B A3B Instruct | 75.93 |
Interactive version: theaggregate.ai/benchmark?slug=helm-arabic-alghafa · How It Works · Data refreshed daily, snapshot 2026-09-08.