Jais 3 Model Cards - SimpleQA Verified (Arabic) - Non-hallucination Rate: leaderboard
Arabic SimpleQA Verified short factual questions, share of answers that do not hallucinate. MBZUAI's Institute of Foundation Models reported this on the Jais 3 model cards (released October 15, 2026), comparing each Jais 3 size with size-matched reference models; this board collects all six cards. The cards state that all reported results use high reasoning effort. Vendor-reported. Metric: Non-hallucination rate (%). Higher is better. Source: https://huggingface.co/collections/IFM/jais-3-6abba9df4f1173ec7ce2929d
Metric: Non-hallucination rate (%). Source: huggingface.co. Saturation forecast: Not forecast. 29 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemma 4 E2B | 84.8 | #1076 |
| 2 | gemma-4-E4B-it | 67 | #566 |
| 3 | MiniMax-M3 | 45.3 | #137 |
| 4 | Qwen 3.5 2B | 18.2 | #1145 |
| 5 | Gemma 4 12B (IT) | 17.1 | #371 |
| 6 | Gemma 4 31B (IT) | 10.5 | #174 |
| 7 | Falcon-H1-34B-Instruct | 9.4 | #666 |
| 8 | Fanar-1-9B-Instruct | 6.6 | #633 |
| 9 | Qwen 3.5 4B | 6.3 | #473 |
| 10 | Qwen 3.8 27B | 5.4 | #104 |
| 11 | Falcon-H1-7B-Instruct | 4.5 | #1043 |
| 12 | Qwen 3.5 9B | 4.4 | #377 |
| 13 | Qwen3.8-Flash-Next | 3.9 | #69 |
| 14 | Muse Glimmer 30B | 3.7 | #178 |
| 15 | Tiny Aya Global | 3.2 | #1322 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=jais-3-model-cards-simpleqa-verified-arabic-non-hallucination-rate · How It Works · Data refreshed daily, snapshot 2026-10-19.