HELM Arabic - Mbzuai Human Translated Arabic Mmlu - Prehistory: leaderboard
Metric: Exact Match (%). Source: crfm.stanford.edu. 41 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.7 | 90.74 |
| 2 | GPT-5.4 (2026-03-05) | 88.58 |
| 3 | Claude Sonnet 4.6 | 87.35 |
| 4 | GPT-4.1 (2025-04-14) | 86.11 |
| 5 | Gemini 2.5 Flash (Thinking) | 85.8 |
| 6 | Gemma 4 31B (IT) | 81.79 |
| 7 | Mistral Large 3 | 81.17 |
| 8 | Qwen 3.5 397B A17B | 80.86 |
| 9 | Qwen 3 Next 80B A3B Instruct | 78.7 |
| 10 | Claude Haiku 4.5 (20251001) | 78.4 |
| 11 | Qwen 2.5 72B Instruct | 76.85 |
| 12 | Llama 4 Maverick Instruct FP8 | 76.85 |
| 13 | DeepSeek V3.1 | 75.62 |
| 14 | Gemini 2.5 Flash Lite (Thinking) | 75 |
| 15 | GPT-4.1 Mini | 74.69 |
Interactive version: theaggregate.ai/benchmark?slug=helm-arabic-mbzuai-human-translated-arabic-mmlu-prehistory · How It Works · Data refreshed daily, snapshot 2026-09-19.