Open Arabic LLM - Aratrust MentalHealth: leaderboard
The "Aratrust MentalHealth" subtask of Open Arabic LLM. A task from the Open Arabic LLM Leaderboard, evaluating models on Arabic-language knowledge, reasoning and understanding across native and translated benchmarks.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 162 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | AceGPT-v2-8B-Chat | 96.05 | #736 |
| 2 | Fanar-1-9B-Instruct | 96.05 | #633 |
| 3 | Qwentile2.5-32B-Instruct | 96.05 | #339 |
| 4 | Awqward2.5-32B-Instruct | 96.05 | #371 |
| 5 | tempmotacilla-cinerea-0308 | 96.05 | #395 |
| 6 | Saka-14B | 96.05 | #344 |
| 7 | Qwen2.5-32B-Instruct-CFT | 96.05 | #375 |
| 8 | Qwen 2.5 14B Instruct | 94.74 | #647 |
| 9 | gemma-3-12B-pt | 94.74 | #686 |
| 10 | gemma-3-27B-pt | 94.74 | #502 |
| 11 | Qwen3-8B-Base | 94.74 | #626 |
| 12 | Qwen 2.5 72B | 94.74 | #499 |
| 13 | jais-adapted-70B-chat | 94.74 | #636 |
| 14 | Qwen 2 72B | 94.74 | #521 |
| 15 | Qwen3-4B-Base | 94.74 | #897 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=open-arabic-llm-aratrust-mentalhealth · How It Works · Data refreshed daily, snapshot 2026-10-20.