UrduMMLU (Urdu Prompt) - Social Sciences: leaderboard
The Social Sciences domain of UrduMMLU (2026), an Urdu MMLU-style benchmark from native question banks and Pakistani examinations: exact-match accuracy in percent on questions from eleven social-science subjects including Pakistan studies, civics and economics, with an Urdu instruction prompt, zero-shot. Higher is better.
Metric: Social Sciences (7,959 questions in 11 subjects, including Pakistan studies, civics and economics) 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: Estimated already saturated. 30 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.5 Flash | 92.25 | #101 |
| 2 | Gemini 3.1 Flash Lite | 90.2 | #364 |
| 3 | GPT-5.4 | 89.47 | #123 |
| 4 | DeepSeek V4 Flash | 88.78 | #258 |
| 5 | Claude Sonnet 4.6 | 87.62 | #161 |
| 6 | Qwen 3.6 35B A3B | 84.19 | #462 |
| 7 | Gemma 4 31B (IT) | 82.12 | #342 |
| 8 | Llama 4 Maverick Instruct | 81.38 | #1133 |
| 9 | GPT-5.4 Mini | 78.54 | #389 |
| 10 | Claude Haiku 4.5 | 77.06 | #568 |
| 11 | Gemma 4 26B A4B (IT) | 75.46 | #549 |
| 12 | Qwen 3.6 27B | 74.63 | #341 |
| 13 | Llama 3.3 70B Instruct | 73.51 | #1357 |
| 14 | Llama 4 Scout Instruct | 72.57 | #1543 |
| 15 | Ministral-3-8B-Instruct-2512 | 62.81 | #1858 |
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-social-sciences · How It Works · Data refreshed daily, snapshot 2026-11-13.