Open Japanese LLM Leaderboard: leaderboard
Japanese language evaluation across 43 tasks including JCommonsenseQA, JMMLU, JSQuAD, MAWPS, and multilingual benchmarks.
Metric: Average 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 | 68.4 | #482 |
| 2 | QwQ 32B-Preview | 67.15 | #651 |
| 3 | Awqward2.5-32B-Instruct | 66.65 | #371 |
| 4 | Llama 3.1 70B Instruct | 66.38 | #561 |
| 5 | qwen2.5-bakeneko-32B-instruct-v2 | 66.35 | #562 |
| 6 | Llama 3.1 Nemotron 70B Instruct | 66.24 | #554 |
| 7 | Qwen2.5-32B-Instruct-abliterated-v2 | 65.61 | #394 |
| 8 | Rombos-LLM-V2.5-Qwen-32B | 65.58 | #367 |
| 9 | Qwen 2.5 32B Instruct | 65.53 | #498 |
| 10 | Qwen2.5-32B-Instruct-CFT | 65.51 | #375 |
| 11 | llm-jp-3.1-13B-instruct4 | 65.25 | #835 |
| 12 | Saka-14B | 65.17 | #344 |
| 13 | Lamarckvergence-14B | 65.11 | #545 |
| 14 | tempmotacilla-cinerea-0308 | 64.65 | #395 |
| 15 | Qwentile2.5-32B-Instruct | 64.57 | #339 |
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-leaderboard · How It Works · Data refreshed daily, snapshot 2026-10-20.