HELM Arabic - Aratrust - Mental Health: leaderboard
Metric: Exact Match (%). Source: crfm.stanford.edu. 41 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.7 | 98.68 |
| 2 | GPT-4.1 Mini | 97.37 |
| 3 | Gemma 4 31B (IT) | 97.37 |
| 4 | DeepSeek V3.1 | 97.37 |
| 5 | Mistral Large 3 | 97.37 |
| 6 | Claude Haiku 4.5 (20251001) | 96.05 |
| 7 | Qwen 3 Next 80B A3B Instruct | 96.05 |
| 8 | Qwen 3.5 397B A17B | 96.05 |
| 9 | Gemini 2.5 Flash Lite (Thinking) | 96.05 |
| 10 | GPT-5.4 (2026-03-05) | 96.05 |
| 11 | Llama 3.3 70B Instruct | 94.74 |
| 12 | Claude Sonnet 4.6 | 94.74 |
| 13 | GPT-4.1 (2025-04-14) | 94.74 |
| 14 | Qwen 3.5 9B | 94.74 |
| 15 | AceGPT-v2-32B-Chat | 94.74 |
Interactive version: theaggregate.ai/benchmark?slug=helm-arabic-aratrust-mental-health · How It Works · Data refreshed daily, snapshot 2026-09-19.