Nejumi 4 - GLP - Knowledge & QA: leaderboard
An aggregate of Nejumi 4's two knowledge sub-scores. General knowledge comes from jaster commonsense, multi-hop and quiz sets plus the MT-Bench STEM questions; specialized knowledge comes from JMMLU, the Japanese MMLU-ProX and a Japanese subset of Humanity's Last Exam. All of it is closed-book, and the board is the mean of the two in percent, higher is better. Source: github.com/wandb/llm-leaderboard.
Metric: Score (%). Source: nejumi.ai. Saturation forecast: Around January 2027. 139 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.6 Sol | 81.54 | #19 |
| 2 | Gemini 3.1 Pro (Preview) | 81.41 | #58 |
| 3 | Claude Fable 5 | 81.34 | #16 |
| 4 | GPT-5.5 (2026-04-23) | 80.96 | #22 |
| 5 | GPT-5.4 (2026-03-05) | 80.63 | #59 |
| 6 | GPT-5.6 Terra | 80.38 | #41 |
| 7 | Claude Opus 5 | 80.01 | #12 |
| 8 | Gemini 3.5 Flash | 79.48 | #60 |
| 9 | Muse Spark Open | 79.38 | #54 |
| 10 | Gemini 3.6 Flash | 79.04 | #51 |
| 11 | Qwen3.8 2.4T A95B | 78.93 | #55 |
| 12 | Claude Opus 4.7 | 78.91 | #49 |
| 13 | Hy4 preview | 78.79 | #38 |
| 14 | Claude Opus 4.8 | 78.72 | #37 |
| 15 | DeepSeek V4.1 Flash | 78.68 | #36 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=nejumi-4-glp-knowledge-qa · How It Works · Data refreshed daily, snapshot 2026-10-20.