HiMed-West - MMLU-Pro Health (Hindi): leaderboard
Hindi translation of MMLU-Pro Health in HiMed-West Bench (HiMed (Jiang et al., 2026)): 818 questions with up to ten options, translated by NLLB-200 with an audited English-Hindi medical lexicon that keeps key terms in parentheses. Scored as zero-shot multiple choice scored with a relaxed answer extractor, averaged over three runs; accuracy in percent, higher is better.
Metric: Accuracy (%; 818 translated MMLU-Pro Health questions, zero-shot multiple choice, mean of three runs). Source: arxiv.org. Saturation forecast: Estimated already saturated. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o | 47 | #741 |
| 2 | Qwen 2.5 14B Instruct | 26 | #1745 |
| 3 | Qwen 2.5 7B Instruct | 20.3 | #2583 |
| 4 | Llama 3.1 8B Instruct | 19.7 | #4277 |
| 5 | aya-23-8B | 14.8 | #6778 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=himed-west-mmlu-pro-health-hindi · How It Works · Data refreshed daily, snapshot 2026-11-12.