HELM Arabic - Alghafa - Multiple Choice Facts Truefalse Balanced Task: leaderboard
Metric: Exact Match (%). Source: crfm.stanford.edu. 41 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Llama 3.3 70B Instruct | 97.33 |
| 2 | AceGPT-v2-32B-Chat | 97.33 |
| 3 | jais-adapted-70B-chat | 96 |
| 4 | Qwen 2.5 72B Instruct | 94.67 |
| 5 | Llama 4 Scout Instruct | 94.67 |
| 6 | Qwen 3.5 397B A17B | 94.67 |
| 7 | jais-family-30B-16k-chat | 94.67 |
| 8 | GPT-4.1 (2025-04-14) | 93.33 |
| 9 | GPT-5.4 (2026-03-05) | 93.33 |
| 10 | Gemma 4 31B (IT) | 92 |
| 11 | DeepSeek V3.1 | 92 |
| 12 | Mistral Large 2 (Nov) Instruct (2411) | 92 |
| 13 | Gemini 2.5 Flash Lite (Thinking) | 92 |
| 14 | Claude Opus 4.7 | 90.67 |
| 15 | Claude Haiku 4.5 (20251001) | 90.67 |
Interactive version: theaggregate.ai/benchmark?slug=helm-arabic-alghafa-multiple-choice-facts-truefalse-balanced-task · How It Works · Data refreshed daily, snapshot 2026-09-19.