Open Japanese LLM - EL: leaderboard
The "EL" 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 | llm-jp-3.1-13B-instruct4 | 76.52 | #835 |
| 2 | llm-jp-3.1-1.8B-instruct4 | 64.94 | #1273 |
| 3 | cogito-v1-preview-qwen-14B | 59.71 | |
| 4 | Qwen2.5-32B-Instruct-CFT | 58.96 | #375 |
| 5 | Qwen 2.5 32B Instruct | 58.94 | #498 |
| 6 | Llama 3 70B Instruct | 58.91 | #634 |
| 7 | Rombos-LLM-V2.5-Qwen-32B | 58.8 | #367 |
| 8 | DeepSeek-R1-Distill-Qwen-32B-Japanese | 58.6 | #638 |
| 9 | Awqward2.5-32B-Instruct | 58.49 | #371 |
| 10 | oxyge1-33B | 58.42 | #690 |
| 11 | Llama-3.3-Swallow-70B-Instruct-v0.4 | 58.38 | #529 |
| 12 | DeepSeek R1 Distill Qwen 14B | 58.24 | #844 |
| 13 | Qwentile2.5-32B-Instruct | 58.05 | #339 |
| 14 | Llama-3.1-Tulu-3-70B | 57.98 | #482 |
| 15 | QwQ 32B-Preview | 57.94 | #651 |
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-el · How It Works · Data refreshed daily, snapshot 2026-10-20.