Open Japanese LLM - Xlsum JA Rouge1: leaderboard
The "Xlsum JA Rouge1" subtask of Open Japanese LLM. A task from the Open Japanese LLM Leaderboard, evaluating models on Japanese-language knowledge, reasoning and understanding.
Metric: Score (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 418 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama-3.1-Tulu-3-70B | 47.5 | #482 |
| 2 | Gemma 2 9B (IT) | 36.16 | #788 |
| 3 | Gemma 2 27B (IT) | 35.88 | #645 |
| 4 | aya-expanse-32B | 35.73 | #692 |
| 5 | Qwen 1.5 32B | 35.64 | #794 |
| 6 | Falcon3-10B-Instruct | 35.03 | #1157 |
| 7 | aya-23-35B | 34.08 | #883 |
| 8 | Qwen 2 57B A14B | 33.43 | #783 |
| 9 | Qwen 2.5 32B | 32.49 | #546 |
| 10 | granite-3.1-2B-instruct | 32.31 | #1385 |
| 11 | Linkbricks-Horizon-AI-Korean-Superb-22B | 31.71 | #708 |
| 12 | qwen2.5-bakeneko-32B-instruct-v2 | 31.59 | #562 |
| 13 | llm-jp-3.1-13B-instruct4 | 31.43 | #835 |
| 14 | Qwen 2 7B | 31 | #1095 |
| 15 | aya-23-8B | 30.82 | #1299 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=open-japanese-llm-xlsum-ja-rouge1 · How It Works · Data refreshed daily, snapshot 2026-10-20.