UrduMMLU (English Prompt): leaderboard
Metric: Overall 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, English instruction prompt (question and options in Urdu). Source: arxiv.org. Saturation forecast: Around December 2026. 30 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.5 Flash | 90.23 |
| 2 | Gemini 3.1 Flash Lite | 84.69 |
| 3 | Claude Sonnet 4.6 | 83.04 |
| 4 | DeepSeek V4 Flash | 82.45 |
| 5 | GPT-5.4 | 80.94 |
| 6 | Gemma 4 31B (IT) | 76.49 |
| 7 | Llama 4 Maverick Instruct | 75.59 |
| 8 | GPT-5.4 Mini | 73.55 |
| 9 | Claude Haiku 4.5 | 71.44 |
| 10 | Qwen 3.6 35B A3B (Non-reasoning) | 69.5 |
| 11 | Gemma 4 26B A4B (IT) | 69.34 |
| 12 | Qwen 3.6 27B | 69.33 |
| 13 | Llama 3.3 70B Instruct | 68.67 |
| 14 | Llama 4 Scout Instruct | 67.93 |
| 15 | Gemma 2 9B (IT) | 55.37 |
Interactive version: theaggregate.ai/benchmark?slug=urdummlu-english-prompt · How It Works · Data refreshed daily, snapshot 2026-09-29.