Open Japanese LLM - CG: leaderboard
The "CG" 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 | Qwen 2.5 Coder 32B Instruct | 68.47 | #528 |
| 2 | QwQ 32B-Preview | 66.06 | #651 |
| 3 | Saka-14B | 64.46 | #344 |
| 4 | Awqward2.5-32B-Instruct | 64.26 | #371 |
| 5 | qwen2.5-bakeneko-32B-instruct-v2 | 63.86 | #562 |
| 6 | Qwen2.5-14B-Instruct-1M | 63.86 | #578 |
| 7 | Lamarckvergence-14B | 63.25 | #545 |
| 8 | Llama-3.1-Tulu-3-70B | 60.64 | #482 |
| 9 | Llama 3.1 70B Instruct | 60.04 | #561 |
| 10 | Qwen2.5-32B-Instruct-abliterated-v2 | 59.44 | #394 |
| 11 | tempmotacilla-cinerea-0308 | 58.43 | #395 |
| 12 | Qwen 2.5 14B | 57.63 | #725 |
| 13 | Llama 3.1 Nemotron 70B Instruct | 55.82 | #554 |
| 14 | Rombos-LLM-V2.5-Qwen-32B | 53.21 | #367 |
| 15 | Qwen2.5-32B-Instruct-CFT | 53.01 | #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-cg · How It Works · Data refreshed daily, snapshot 2026-10-20.