UrduMMLU (Urdu Prompt) - Humanities: leaderboard
The Humanities domain of UrduMMLU (2026), an Urdu MMLU-style benchmark from native question banks and Pakistani examinations: exact-match accuracy in percent on ethics, fine arts, Islamic studies and Urdu grammar, language and literature questions, with an Urdu instruction prompt, zero-shot. Higher is better.
Metric: Humanities (10,982 questions: ethics, fine arts, Islamic studies, Urdu grammar, language and literature) exact-match accuracy (%) on UrduMMLU's Urdu multiple-choice questions from Pakistani MCQ banks and public examinations, zero-shot, answer key parsed from the generation (unparsable, malformed or error outputs count as wrong); temperature 0 where available, 4,096 output tokens, Urdu instruction prompt. Source: arxiv.org. Saturation forecast: Around February 2027. 30 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.5 Flash | 85.49 | #101 |
| 2 | Gemini 3.1 Flash Lite | 74.57 | #364 |
| 3 | GPT-5.4 | 74.29 | #123 |
| 4 | Claude Sonnet 4.6 | 72.87 | #161 |
| 5 | DeepSeek V4 Flash | 67.3 | #258 |
| 6 | Llama 4 Maverick Instruct | 63.41 | #1133 |
| 7 | Gemma 4 31B (IT) | 63.38 | #342 |
| 8 | GPT-5.4 Mini | 62.51 | #389 |
| 9 | Claude Haiku 4.5 | 59.42 | #568 |
| 10 | Qwen 3.6 35B A3B | 57.98 | #462 |
| 11 | Gemma 4 26B A4B (IT) | 57.87 | #549 |
| 12 | Llama 4 Scout Instruct | 56.69 | #1543 |
| 13 | Llama 3.3 70B Instruct | 56.24 | #1357 |
| 14 | Qwen 3.6 27B | 55.86 | #341 |
| 15 | Gemma 2 9B (IT) | 48.21 | #2262 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=urdummlu-urdu-prompt-humanities · How It Works · Data refreshed daily, snapshot 2026-11-13.