Open Japanese LLM - HE: leaderboard
The "HE" 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 | cogito-v1-preview-qwen-14B | 79.2 | |
| 2 | ABEJA-Qwen2.5-32B-Japanese-v1.0 | 78.81 | #503 |
| 3 | Qwentile2.5-32B-Instruct | 78.64 | #339 |
| 4 | Rombos-LLM-V2.5-Qwen-32B | 77.69 | #367 |
| 5 | Awqward2.5-32B-Instruct | 77.68 | #371 |
| 6 | oxyge1-33B | 77.61 | #690 |
| 7 | Qwen 2.5 32B Instruct | 77.57 | #498 |
| 8 | QwQ 32B-Preview | 77.56 | #651 |
| 9 | Qwen2.5-32B-Instruct-CFT | 77.54 | #375 |
| 10 | Qwen2.5-32B-Instruct-abliterated-v2 | 77.41 | #394 |
| 11 | Llama 3.3 70B Instruct | 77.14 | #534 |
| 12 | Qwen 2.5 32B | 77.14 | #546 |
| 13 | Llama 3.1 70B Instruct | 76.78 | #561 |
| 14 | Llama 3.1 Nemotron 70B Instruct | 76.73 | #554 |
| 15 | EZO-Qwen2.5-32B-Instruct | 76.33 | #471 |
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-he · How It Works · Data refreshed daily, snapshot 2026-10-20.