Nejumi 4 - ALT - Ethics & Morality: leaderboard
Japanese commonsense morality items from the jaster battery, where the model reads a sentence describing an action and answers with a single label saying whether it is morally acceptable. Only the two-shot jaster run feeds this board, graded by exact match against the dataset label and reported as a percentage where 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 | Claude Opus 4.7 | 99 | #49 |
| 2 | Gemini 3.5 Flash | 99 | #60 |
| 3 | Claude Opus 4.6 | 98 | #63 |
| 4 | Claude Opus 4.8 | 98 | #37 |
| 5 | Gemini 3 Flash (Preview) | 98 | #82 |
| 6 | Gemini 3.1 Flash Lite (Preview) | 98 | #194 |
| 7 | Claude Fable 5 | 98 | #16 |
| 8 | Gemini 3.6 Flash | 98 | #51 |
| 9 | Gemini 2.5 Pro | 97 | #152 |
| 10 | Gemini 3.1 Pro (Preview) | 97 | #58 |
| 11 | Gemini 3 Pro (Preview) | 97 | #66 |
| 12 | Qwen 3.5 397B A17B | 97 | #146 |
| 13 | Gemma 4 31B (IT) | 97 | #176 |
| 14 | GPT-4o Mini (2024-07-18) | 97 | #558 |
| 15 | Claude 3.7 Sonnet | 97 | #249 |
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-alt-ethics-morality · How It Works · Data refreshed daily, snapshot 2026-10-20.