Open Japanese LLM - ALT E TO J Bleu JA: leaderboard
The "ALT E TO J Bleu JA" 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: Around 2032. 418 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | llm-jp-3.1-13B-instruct4 | 15.7 | #835 |
| 2 | Llama-3.3-Swallow-70B-Instruct-v0.4 | 14.94 | #529 |
| 3 | llm-jp-3.1-1.8B-instruct4 | 14.78 | #1273 |
| 4 | aya-expanse-32B | 14.77 | #692 |
| 5 | Llama 3 70B | 13.97 | #631 |
| 6 | Llama 3.1 Nemotron 70B Instruct | 13.92 | #554 |
| 7 | ABEJA-Qwen2.5-32B-Japanese-v1.0 | 13.88 | #503 |
| 8 | Llama 3.1 70B | 13.75 | #592 |
| 9 | Llama-3.1-Tulu-3-70B | 13.73 | #482 |
| 10 | Llama 3.1 70B Instruct | 13.71 | #561 |
| 11 | Qwen2.5-32B-Instruct-abliterated-v2 | 13.6 | #394 |
| 12 | Llama 3.3 70B Instruct | 13.59 | #534 |
| 13 | llm-jp-3-13B | 13.53 | #1219 |
| 14 | aya-23-35B | 13.52 | #883 |
| 15 | Qwen2.5-32B-Instruct-CFT | 13.32 | #375 |
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-alt-e-to-j-bleu-ja · How It Works · Data refreshed daily, snapshot 2026-10-20.