HELM Arabic - Arabic Exams - Physics: leaderboard
HELM Arabic (release v2.2.0) Arabic EXAMS, Physics questions: Arabic high-school examination multiple-choice questions from the Arabic subset of EXAMS (OALL mirror of the AceGPT version); 42 questions, exact match. Percent correct, 0-100, higher is better.
Metric: Exact Match (%). Source: crfm.stanford.edu. Saturation forecast: Around February 2027. 41 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.7 | 73.81 | #51 |
| 2 | Llama 3.3 70B Instruct | 69.05 | #541 |
| 3 | GPT-4.1 (2025-04-14) | 69.05 | #257 |
| 4 | GPT-5.1 (2025-11-13) | 69.05 | #107 |
| 5 | Qwen 3.5 397B A17B | 66.67 | #154 |
| 6 | Gemma 4 31B (IT) | 66.67 | #178 |
| 7 | Gemini 2.5 Flash (Non-reasoning) | 66.67 | #255 (Gemini 2.5 Flash) |
| 8 | Gemini 2.5 Flash Lite (Non-reasoning) | 66.67 | #432 (Gemini 2.5 Flash Lite) |
| 9 | GPT-5.4 (2026-03-05) | 64.29 | #63 |
| 10 | Claude Sonnet 4.6 | 61.9 | #94 |
| 11 | Llama 4 Scout Instruct | 61.9 | #547 |
| 12 | Llama 4 Maverick Instruct FP8 | 61.9 | #424 |
| 13 | Qwen 3.5 9B | 59.52 | #391 |
| 14 | Claude Haiku 4.5 (20251001) | 59.52 | #264 |
| 15 | AceGPT-v2-70B-Chat | 59.52 | #421 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=helm-arabic-arabic-exams-physics · How It Works · Data refreshed daily, snapshot 2026-11-05.