Open Japanese LLM - NLI: leaderboard
The "NLI" 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 | 96.5 | #835 |
| 2 | llm-jp-3.1-1.8B-instruct4 | 92.44 | #1273 |
| 3 | ABEJA-Qwen2.5-32B-Japanese-v1.0 | 81.46 | #503 |
| 4 | Qwen2.5-32B-Instruct-abliterated-v2 | 81.38 | #394 |
| 5 | qwen2.5-bakeneko-32B-instruct-v2 | 81.3 | #562 |
| 6 | Qwen2.5-32B-Instruct-CFT | 81.09 | #375 |
| 7 | Qwen 2.5 32B Instruct | 81.06 | #498 |
| 8 | Rombos-LLM-V2.5-Qwen-32B | 81.05 | #367 |
| 9 | oxyge1-33B | 80.97 | #690 |
| 10 | EZO-Qwen2.5-32B-Instruct | 80.95 | #471 |
| 11 | Awqward2.5-32B-Instruct | 80.92 | #371 |
| 12 | Qwentile2.5-32B-Instruct | 80.59 | #339 |
| 13 | Llama 3.3 70B Instruct | 80.16 | #534 |
| 14 | cogito-v1-preview-qwen-14B | 79.12 | |
| 15 | QwQ 32B-Preview | 79.12 | #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-nli · How It Works · Data refreshed daily, snapshot 2026-10-20.