HELM Arabic - Alghafa - Multiple Choice Facts Truefalse Balanced Task: leaderboard
HELM Arabic (release v2.2.0) AlGhafa Multiple Choice Facts Truefalse Balanced Task task: true/false judgments of factual statements, balanced between the two answers, from the native AlGhafa Arabic multiple-choice benchmark; 75 items, exact match. Percent correct, 0-100, higher is better.
Metric: Exact Match (%). Source: crfm.stanford.edu. Saturation forecast: Estimated already saturated. 41 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 3.3 70B Instruct | 97.33 | #541 |
| 2 | AceGPT-v2-32B-Chat | 97.33 | #518 |
| 3 | jais-adapted-70B-chat | 96 | #639 |
| 4 | Qwen 2.5 72B Instruct | 94.67 | #469 |
| 5 | Qwen 3.5 397B A17B | 94.67 | #154 |
| 6 | Llama 4 Scout Instruct | 94.67 | #547 |
| 7 | jais-family-30B-16k-chat | 94.67 | #758 |
| 8 | GPT-4.1 (2025-04-14) | 93.33 | #257 |
| 9 | GPT-5.4 (2026-03-05) | 93.33 | #63 |
| 10 | GPT-5.4 Mini (2026-03-17) | 93.33 | #180 |
| 11 | Gemma 4 31B (IT) | 92 | #178 |
| 12 | DeepSeek V3.1 | 92 | #275 |
| 13 | GPT-5.1 (2025-11-13) | 92 | #107 |
| 14 | Mistral Large 2 (Nov) Instruct (2411) | 92 | #480 |
| 15 | Gemini 2.5 Flash Lite (Non-reasoning) | 92 | #432 (Gemini 2.5 Flash Lite) |
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-alghafa-multiple-choice-facts-truefalse-balanced-task · How It Works · Data refreshed daily, snapshot 2026-11-05.