Open Japanese LLM - QA: leaderboard
The "QA" 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-3.3-Swallow-70B-Instruct-v0.4 | 72.74 | #529 |
| 2 | llm-jp-3.1-13B-instruct4 | 69.31 | #835 |
| 3 | llm-jp-3-13B | 67.21 | #1219 |
| 4 | ABEJA-Qwen2.5-32B-Japanese-v1.0 | 66.79 | #503 |
| 5 | Llama-3.1-Tulu-3-70B | 66.05 | #482 |
| 6 | aya-expanse-32B | 64.95 | #692 |
| 7 | Llama 3.1 70B | 64.87 | #592 |
| 8 | Llama 3 70B | 64.65 | #631 |
| 9 | Llama 3.1 70B Instruct | 64.56 | #561 |
| 10 | Llama 3.1 Nemotron 70B Instruct | 64.53 | #554 |
| 11 | Llama 3.3 70B Instruct | 64.12 | #534 |
| 12 | Llama 3 70B Instruct | 61.25 | #634 |
| 13 | aya-23-35B | 59.91 | #883 |
| 14 | Gemma 2 27B (IT) | 59.32 | #645 |
| 15 | Qwentile2.5-32B-Instruct | 58.26 | #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-qa · How It Works · Data refreshed daily, snapshot 2026-10-20.