Swallow - Pre-trained Japanese - NIILC: leaderboard
From the Swallow LLM Leaderboard (Institute of Science Tokyo) pre-trained Japanese board, which evaluates base models without post-training. NIILC open-ended encyclopedic question answering, 4-shot; metric is character-level F1. Upstream publishes 0-1 scores; shown as 0-100, higher is better.
Metric: Character F1 (%). Source: swallow-llm.github.io. Saturation forecast: Estimated already saturated. 50 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | sarashina2-70B | 66.8 | #586 |
| 2 | sarashina2-13B | 66.1 | #898 |
| 3 | plamo-2-8B | 65.5 | #839 |
| 4 | llm-jp-3-13B | 64.9 | #1176 |
| 5 | sarashina2.2-3B | 64.2 | #766 |
| 6 | sarashina2-7B | 63.4 | #1134 |
| 7 | gemma-3-27B-pt | 62.7 | #505 |
| 8 | Qwen 2.5 72B | 61.9 | #512 |
| 9 | Llama 4 Scout Base | 61.6 | #708 |
| 10 | Llama 3.1 70B | 60.3 | #588 |
| 11 | llm-jp-3-7.2B | 60.1 | #1222 |
| 12 | Gemma 2 27B | 57.3 | #803 |
| 13 | gemma-3-12B-pt | 56.9 | #686 |
| 14 | llm-jp-3-3.7B | 54.1 | #1407 |
| 15 | Qwen 2.5 32B | 53.8 | #544 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=swallow-pre-trained-japanese-niilc · How It Works · Data refreshed daily, snapshot 2026-11-05.