Jais 3 Model Cards - Arabic IFBench (IFM): leaderboard
IFM's Arabic IFBench instruction following, prompt-level strict accuracy (benchmark not yet released). 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: Prompt-level strict accuracy (%). Higher is better. Source: https://huggingface.co/collections/IFM/jais-3-6abba9df4f1173ec7ce2929d
Metric: Prompt-level strict accuracy (%). Source: huggingface.co. Saturation forecast: Around January 2027. 29 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Muse Glimmer 30B | 84.2 | #178 |
| 2 | Qwen3.8-Flash-Next | 81.7 | #69 |
| 3 | Qwen 3.8 27B | 81.3 | #104 |
| 4 | Gemma 4 31B (IT) | 80.8 | #174 |
| 5 | Nemotron 3 Ultra | 78.9 | #234 |
| 6 | Gemma 4 12B (IT) | 75.3 | #371 |
| 7 | Qwen 3.5 397B A17B | 75.1 | #144 |
| 8 | Qwen 3.5 9B | 58.3 | #377 |
| 9 | MiniMax-M3 | 53.8 | #137 |
| 10 | Qwen 3.5 4B | 53.7 | #473 |
| 11 | gemma-4-E4B-it | 42.6 | #566 |
| 12 | Gemma 4 E2B | 37.6 | #1076 |
| 13 | Qwen 3.5 2B | 33.1 | #1145 |
| 14 | Falcon-H1-34B-Instruct | 32.2 | #666 |
| 15 | Falcon-H1-7B-Instruct | 26.2 | #1043 |
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-arabic-ifbench-ifm · How It Works · Data refreshed daily, snapshot 2026-10-19.