Hala 1.2B: benchmark results
Provider: Other.
Aggregate Rating 1305 ± 41, rank #8574 of 10552 rated models, from 28 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Jais 3 Model Cards - AraGen v3 - Conciseness | 14.4 | 3C3H Conciseness (%) | 41.1 |
| Jais 3 Model Cards - SimpleQA Verified (Arabic) - F-score | 3.1 | F-score (%) | 25 |
| Jais 3 Model Cards - SimpleQA Verified (Arabic) - Non-hallucination Rate | 3.2 | Non-hallucination rate (%) | 16.1 |
| Jais 3 Model Cards - AraGen v3 (3C3H) | 25 | 3C3H Score (%) | 14.3 |
| Jais 3 Model Cards - IslamicEval 2025-1B (IFM-audited) | 35.5 | Accuracy (%) | 10.7 |
| Jais 3 Model Cards - Alyah (IFM-audited) | 53.2 | Accuracy (%) | 5.4 |
| Jais 3 Model Cards - AraSTEM (IFM-audited) | 48.3 | Accuracy (%) | 3.6 |
| Jais 3 Model Cards - ArabCulture (IFM-audited) | 60.6 | Accuracy (%) | 3.6 |
| Jais 3 Model Cards - Arabic Cuisine (IFM) | 49.5 | Accuracy (%) | 3.6 |
| Jais 3 Model Cards - Arabic GSM8K (IFM-audited) | 57.3 | Accuracy (%) | 3.6 |
| Jais 3 Model Cards - Arabic IFEval | 44.1 | Accuracy (%) | 3.6 |
| Jais 3 Model Cards - Arabic Linguistics - Diacritization (IFM) | 32.9 | Accuracy (%) | 3.6 |
Interactive version: theaggregate.ai/model?slug=hala-1-2b · How It Works · Data refreshed daily, snapshot 2026-11-16.