Nejumi 4 - GLP - Basic Language: leaderboard
An aggregate of Nejumi 4's two foundational language sub-scores. Semantic analysis covers Japanese natural language inference sets, namely JNLI, JaNLI, JSeM, JSICK and JAMP; syntactic analysis covers grammatical acceptability on JCoLA in-domain and out-of-domain and on JBLiMP. 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: Estimated already saturated. 139 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 (2026-04-23) | 87.8 | #22 |
| 2 | Gemini 3 Flash (Preview) | 87.62 | #82 |
| 3 | GPT-5.6 Sol | 87 | #19 |
| 4 | GPT-5.6 Terra | 86.95 | #41 |
| 5 | Gemma 4 31B (IT) | 86.82 | #176 |
| 6 | Claude Fable 5 | 86.68 | #16 |
| 7 | Qwen3.8 2.4T A95B | 86.68 | #55 |
| 8 | Claude Opus 4.6 | 86.65 | #63 |
| 9 | GPT-5.1 (2025-11-13) | 86.27 | #101 |
| 10 | Gemini 3 Pro (Preview) | 86.23 | #66 |
| 11 | Gemini 3.5 Flash | 86.22 | #60 |
| 12 | Claude Sonnet 4.6 | 86.13 | #90 |
| 13 | Gemini 3.6 Flash | 86.07 | #51 |
| 14 | Gemini 2.5 Pro | 85.95 | #152 |
| 15 | GPT-5.6 Luna | 85.92 | #68 |
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-basic-language · How It Works · Data refreshed daily, snapshot 2026-10-20.