Nejumi 4 - GLP - Semantic Analysis: leaderboard
Japanese natural language inference items from the jaster harness, drawn from JNLI, JaNLI, JSeM, JSICK and JAMP, where the model labels the relation between a premise and a hypothesis. Each dataset is the mean of its 0-shot and 2-shot runs graded by exact match, and the board averages them in percent where 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 | Claude Opus 5 | 87.1 | #12 |
| 2 | Kimi K3 | 86.4 | #30 |
| 3 | Claude Opus 4.8 | 86.2 | #37 |
| 4 | Qwen 3.6 Max Preview | 86.1 | #85 |
| 5 | GPT-5.6 Sol | 86 | #19 |
| 6 | GPT-5.6 Terra | 85.9 | #41 |
| 7 | Gemini 3.1 Pro (Preview) | 85.8 | #58 |
| 8 | Claude Opus 4.6 | 85.8 | #63 |
| 9 | Qwen3.8 2.4T A95B | 85.7 | #55 |
| 10 | GPT-5.5 (2026-04-23) | 85.6 | #22 |
| 11 | Ornith-1.5-397B | 85.6 | |
| 12 | Gemini 3 Flash (Preview) | 85.4 | #82 |
| 13 | Gemini 3.6 Flash | 85.3 | #51 |
| 14 | Claude Sonnet 4.6 | 85.1 | #90 |
| 15 | Claude Opus 4.5 | 85 | #83 |
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-semantic-analysis · How It Works · Data refreshed daily, snapshot 2026-10-20.