Swallow - Pre-trained Japanese - JCommonsenseQA: leaderboard
Metric: Accuracy (%). Source: swallow-llm.github.io. 50 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 2.5 72B | 97.2 |
| 2 | Qwen 2.5 32B | 96.1 |
| 3 | Llama 4 Scout Base | 95.8 |
| 4 | Qwen 2.5 14B | 95.8 |
| 5 | Qwen3-14B-Base | 95.6 |
| 6 | Llama 3.1 70B | 94.6 |
| 7 | gemma-3-27B-pt | 94.4 |
| 8 | Gemma 2 27B | 93.6 |
| 9 | sarashina2-70B | 92.9 |
| 10 | Qwen3-8B-Base | 92.7 |
| 11 | Qwen3-30B-A3B-Base | 92.7 |
| 12 | Qwen 2.5 7B | 92.4 |
| 13 | sarashina2.2-3B | 91.1 |
| 14 | Qwen3-4B-Base | 91 |
| 15 | plamo-2-8B | 90.9 |
Interactive version: theaggregate.ai/benchmark?slug=swallow-pre-trained-japanese-jcommonsenseqa · How It Works · Data refreshed daily, snapshot 2026-09-19.