HELM Arabic - Alghafa - Multiple Choice Rating Sentiment No Neutral Task: leaderboard
Metric: Exact Match (%). Source: crfm.stanford.edu. 41 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4.6 | 86.3 |
| 2 | Claude Opus 4.7 | 86 |
| 3 | GPT-4.1 (2025-04-14) | 84.4 |
| 4 | Llama 4 Maverick Instruct FP8 | 84.1 |
| 5 | Llama 4 Scout Instruct | 82.8 |
| 6 | Llama 3.3 70B Instruct | 82.4 |
| 7 | jais-adapted-7B-chat | 82 |
| 8 | Gemini 2.5 Flash Lite (Thinking) | 81.7 |
| 9 | Claude Haiku 4.5 (20251001) | 81.5 |
| 10 | Mistral Large 2 (Nov) Instruct (2411) | 81.2 |
| 11 | ALLaM-7B-Instruct-preview | 81 |
| 12 | jais-adapted-70B-chat | 80.3 |
| 13 | Gemma 4 31B (IT) | 79.9 |
| 14 | Qwen 3.5 397B A17B | 79.5 |
| 15 | AceGPT-v2-32B-Chat | 79.5 |
Interactive version: theaggregate.ai/benchmark?slug=helm-arabic-alghafa-multiple-choice-rating-sentiment-no-neutral-task · How It Works · Data refreshed daily, snapshot 2026-09-19.