IndicKLAR (Sindhi): leaderboard
Metric: Accuracy (%; relaxed prefix match of the answer to 2,619 factual queries in Sindhi script, machine-translated, not speaker-verified; direct 3-shot prompting with English examples, greedy decoding). Source: arxiv.org. Saturation forecast: Estimated already saturated. 9 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemma 3 12B (IT) | 54 |
| 2 | Qwen 2.5 14B Instruct | 50.9 |
| 3 | Llama 3.1 8B Instruct | 47.8 |
| 4 | Gemma 3 4B (IT) | 41.6 |
| 5 | Qwen 2.5 7B Instruct | 36.9 |
| 6 | Llama 3.2 3B Instruct | 24.9 |
| 7 | Qwen 2.5 1.5B Instruct | 18.8 |
| 8 | Llama 3.2 1B Instruct | 17.2 |
| 9 | Gemma 3 1B (IT) | 12.9 |
Interactive version: theaggregate.ai/benchmark?slug=indicklar-sindhi · How It Works · Data refreshed daily, snapshot 2026-09-26.