IndicIFEval-Trans - Urdu: leaderboard
Metric: Prompt-level loose accuracy (%, 0-100) on the Urdu prompts; IndicIFEval-Trans: IFEval prompts translated with Sarvam-Translate, localized and verified by native speakers; the 321 prompts valid in every language; IFEval-style verifiable instructions checked by the IFEval rules with Indic NLP Library tokenization and sentence splitting; prompt-level loose accuracy (a prompt counts when all of its instructions are followed, after minor formatting relaxations); greedy decoding through the LM Evaluation Harness; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 22 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro | 89.1 | #77 |
| 2 | GPT-5 | 87.5 | #91 |
| 3 | Gemini 3 Flash | 85.4 | #93 |
| 4 | GPT-5 Mini (2025-08-07) | 81.3 | #165 |
| 5 | Gemma 3 12B (IT) | 77.9 | #655 |
| 6 | Gemma 3 27B (IT) | 77.6 | #509 |
| 7 | Llama 4 Scout Instruct | 76.6 | #526 |
| 8 | Llama 3.3 70B Instruct | 76.3 | #520 |
| 9 | Qwen 3 32B | 75.4 | #424 |
| 10 | Llama 3.1 70B Instruct | 74.1 | #548 |
| 11 | Qwen 3 14B | 70.4 | #524 |
| 12 | Gemma 3 4B (IT) | 67.3 | #971 |
| 13 | Qwen 3 8B | 64.8 | #667 |
| 14 | Qwen 3 4B | 59.8 | #823 |
| 15 | aya-expanse-32B | 57.6 | #680 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=indicifeval-trans-urdu · How It Works · Data refreshed daily, snapshot 2026-10-11.